Monitoring the Balancing of a Pack

WheezyRider

Esteemed Pedelecer
Apr 20, 2020
1,690
938
Been meaning to post this for some time but not got around to it so here goes while I have a moment spare. I think it's worth looking at as the results are quite interesting.

I posted this back in November last year:

Here is a dismantled Yose power pack, generic Chinese cells, crappy 20A BMS. Bought in 2019, charged only to 41V since June 2020, so 2.5 years of 41V charging, hundreds of cycles and thrashed quite hard, up to 28A cont for many of those cycles. In that time it has never had a charge cycle above 41V.


49657



The BMS has failed after the abuse it has had and I have been planning to replace it for a long time with a better one. So what do the cell bank voltages look like?

Cell Bank / Voltage (V)
1 ............ 3.677
2 ............ 3.678
3 ............ 3.678
4 ............ 3.678
5 ............ 3.677
6 ............ 3.661
7 ............ 3.676
8 ............ 3.678
9 ............ 3.679
10 .......... 3.678

Overall pack voltage 36.762V

So after all that abuse and hundreds of cycles, the maximum variation is 17 mV, mostly because just one bank is slightly lower, the rest are within a couple of mV. I expect the variation to increase slightly once the pack is charged - this is what I find with my homebuilt pack I added monitoring ports to.

I will charge it up with a new BMS and see. However, this variation is unlikely to exceed 20 to 30mV. With this pack I never knew what the bank voltages were from new, so it is quite possible bank 6 has always been like this. I will charge to 41V initially and then to 42V with the new BMS and see what happens to bank 6, will it balance? We'll see.

This quite generic pack was abused and not balanced for 2.5 years and hundreds of cycles and yet it hasn't gone seriously out of balance. This shows that for normal use, charging to 41V for a significant number of cycles, and then balancing at 42V every now and then does not mean a pack's BMS has to be replaced with one designed for 41V charging.


So I connected it up to a new BMS of a similar type to the old one, but with a 30A rating rather than a 20A rating. Then I applied 41V until the voltage did not increase any longer and then the pack was left over the weekend to settle:

Cell Bank / Voltage (V)
1 ............ 4.0924
2 ............ 4.0917
3 ............ 4.0924
4 ............ 4.0922
5 ............ 4.0917
6 ............ 4.0753
7 ............ 4.0909
8 ............ 4.0915
9 ............ 4.0929
10 .......... 4.0920

Overall pack voltage 40.903 V

So by charging to 41V, the maximum difference is still only ~18mV.

The question is, would balancing occur if the charging voltage was increased to 41.5V? Also I was interested to see how much energy was being sacrificed by charging to 41V. So I applied 41.5V to the pack. This consumed 0.9 Ah for this 15Ah pack. Estimating losses during charging, this probably means ~0.7 to 0.8 Ah stored. The pack was allowed to rest and the measured:

Cell Bank / Voltage (V)
1 ............ 4.1417
2 ............ 4.1403
3 ............ 4.1414
4 ............ 4.1413
5 ............ 4.1401
6 ............ 4.1146
7 ............ 4.1384
8 ............ 4.1403
9 ............ 4.1428
10 .......... 4.1411

What is interesting is that the maximum difference has been increased by going above 41V charging to 28.2mV. No sign of any balancing occurring. The next step was to charge to 41.6V. A further 0.156Ah was applied. The voltages were:

Cell Bank / Voltage (V)
1 ............ 4.1625
2 ............ 4.1606
3 ............ 4.1617
4 ............ 4.1617
5 ............ 4.1600
6 ............ 4.1290
7 ............ 4.1582
8 ............ 4.1603
9 ............ 4.1634
10 .......... 4.1613

Overall pack voltage 41.581V

Now the maximum difference has increased to 34.4mV. What is also interesting is that some of the cells are higher than 1/10 of the applied charging voltage of 41.592V. Still no sign of any balancing going on. So next step was 41.7V. This should get interesting as this is the voltage the old Yose Power SANS chargers used to be set to. A further 0.067Ah was applied, so energy input is dropping off.

Cell Bank / Voltage (V)
1 ............ 4.1744
2 ............ 4.1725
3 ............ 4.1736
4 ............ 4.1737
5 ............ 4.1720
6 ............ 4.1381
7 ............ 4.1701
8 ............ 4.1716
9 ............ 4.1731
10 .......... 4.1736

Still no sign of balancing, bank 6 was still low, even after several hours. The current draw from the charger was 0.7 mA, so I left it on over night to see if it would balance by morning:

Cell Bank / Voltage (V)
1 ............ 4.1764
2 ............ 4.1763
3 ............ 4.1757
4 ............ 4.1781
5 ............ 4.1777
6 ............ 4.1416
7 ............ 4.1718
8 ............ 4.1685
9 ............ 4.1763
10 .......... 4.1764

The maximum difference was now 36.5 mV, so no balancing. Hence, the SANS power pack that came with the battery at 41.7 V would not have balanced the battery.

I then tried 41.8V. Something seemed to start happening. The power input went up to 0.183Ah. I disconnected after running for several hours and then on the Monday morning I measured the voltages:


Cell Bank / Voltage (V)
1 ............ 4.1738
2 ............ 4.1766
3 ............ 4.1721
4 ............ 4.1763
5 ............ 4.1791
6 ............ 4.1551
7 ............ 4.1693
8 ............ 4.1658
9 ............ 4.1657
10 .......... 4.1740

Overall pack voltage 41.705V

So the maximum difference has decreased to 24mV. It seems as if some balancing has occurred, but not enough. The next step was to go to 41.9 V. Charging was performed at 41.903V, 0.125Ah was applied. I was monitoring the voltages closely and I was getting a bit concerned about bank 10 as this was over 4.2V per cell.

Cell Bank / Voltage (V)
1 ............ 4.1904
2 ............ 4.1933
3 ............ 4.1888
4 ............ 4.1931
5 ............ 4.1964
6 ............ 4.1837
7 ............ 4.1859
8 ............ 4.1820
9 ............ 4.1822
10 .......... 4.2018

Overall pack voltage 41.898V

Finally, bank 6 is no longer the lowest bank, although the voltage difference between bank 6 and bank 10 is still 18 mV.

I left it charging for several hours then switched it off to rest for a couple of days. I was getting worried as this was pushing one bank over what I was happy with.

Cell Bank / Voltage (V)
1 ............ 4.1741
2 ............ 4.1769
3 ............ 4.1721
4 ............ 4.1767
5 ............ 4.1795
6 ............ 4.1601
7 ............ 4.1664
8 ............ 4.1651
9 ............ 4.1656
10 .......... 4.1744

Overall pack voltage 41.711V

So after relaxing for a while, bank 6 is the lowest value again and the voltage difference is 19.1mV, which is about the same (or a little worse) as what I started with after the pack hadn't been balanced for years.

The things to take away from this exercise:

Balancing did not start until over 4.18V per cell.

There is negligible additional capacity above 4.15V/cell and only about 5 to 6% between 4.1V and 4.15V/cell

The balancing wasn't all that effective compared to no balance at all, the cells seemed to drop back to where they were comfortable after charging. It is also a question how accurately each balancing module in the BMS can measure and hence balance voltage.

It's easy for a good bank to be driven over 4.2V when the BMS tries to balance - even if the overall applied voltage is less than 4.2V/cell, it can't bleed away enough current. The bigger the difference in the banks, the bigger the problem this will be. So if charging to 4.2V every time, the good banks are going to get a lot of stress. If the charger is over 4.2V/cell I imagine the voltage on the best banks will approach harmful levels and drastically shorten life.

It seems to me the best way of making sure a pack stays in balance is to make sure the cells are matched properly in the first place. I think pack manufacturers should print the bank voltages, capacities and internal resistances on the pack when it is new to show that it has been constructed properly.
 

guerney

Esteemed Pedelecer
Sep 7, 2021
11,355
3,222
It seems to me the best way of making sure a pack stays in balance is to make sure the cells are matched properly in the first place. I think pack manufacturers should print the bank voltages, capacities and internal resistances on the pack when it is new to show that it has been constructed properly.
I don't think that'll ever happen - it makes more financial sense for battery manufacturers (if the there was sufficient demand) if a new generation of smarter ebike battery BMSs was mass produced at similarly low cost as the cheap BMSs you've tried, which monitored each cell continuously to calculate the overall optimum safe balancing voltage of the battery pack, to be built into their battery packs made ad-hoc using cheap mass produced cells of varying capabilities. Which also probably won't happen. There needs to be a strong business case (or regulation) for manufacturers to make batteries safer and longer lived than they are presently - ebike batteries containing dumb BMSs are selling very well.
 
Last edited:

saneagle

Esteemed Pedelecer
Oct 10, 2010
6,791
3,132
Telford
Been meaning to post this for some time but not got around to it so here goes while I have a moment spare. I think it's worth looking at as the results are quite interesting.

I posted this back in November last year:

Here is a dismantled Yose power pack, generic Chinese cells, crappy 20A BMS. Bought in 2019, charged only to 41V since June 2020, so 2.5 years of 41V charging, hundreds of cycles and thrashed quite hard, up to 28A cont for many of those cycles. In that time it has never had a charge cycle above 41V.


49657



The BMS has failed after the abuse it has had and I have been planning to replace it for a long time with a better one. So what do the cell bank voltages look like?

Cell Bank / Voltage (V)
1 ............ 3.677
2 ............ 3.678
3 ............ 3.678
4 ............ 3.678
5 ............ 3.677
6 ............ 3.661
7 ............ 3.676
8 ............ 3.678
9 ............ 3.679
10 .......... 3.678

Overall pack voltage 36.762V

So after all that abuse and hundreds of cycles, the maximum variation is 17 mV, mostly because just one bank is slightly lower, the rest are within a couple of mV. I expect the variation to increase slightly once the pack is charged - this is what I find with my homebuilt pack I added monitoring ports to.

I will charge it up with a new BMS and see. However, this variation is unlikely to exceed 20 to 30mV. With this pack I never knew what the bank voltages were from new, so it is quite possible bank 6 has always been like this. I will charge to 41V initially and then to 42V with the new BMS and see what happens to bank 6, will it balance? We'll see.

This quite generic pack was abused and not balanced for 2.5 years and hundreds of cycles and yet it hasn't gone seriously out of balance. This shows that for normal use, charging to 41V for a significant number of cycles, and then balancing at 42V every now and then does not mean a pack's BMS has to be replaced with one designed for 41V charging.


So I connected it up to a new BMS of a similar type to the old one, but with a 30A rating rather than a 20A rating. Then I applied 41V until the voltage did not increase any longer and then the pack was left over the weekend to settle:

Cell Bank / Voltage (V)
1 ............ 4.0924
2 ............ 4.0917
3 ............ 4.0924
4 ............ 4.0922
5 ............ 4.0917
6 ............ 4.0753
7 ............ 4.0909
8 ............ 4.0915
9 ............ 4.0929
10 .......... 4.0920

Overall pack voltage 40.903 V

So by charging to 41V, the maximum difference is still only ~18mV.

The question is, would balancing occur if the charging voltage was increased to 41.5V? Also I was interested to see how much energy was being sacrificed by charging to 41V. So I applied 41.5V to the pack. This consumed 0.9 Ah for this 15Ah pack. Estimating losses during charging, this probably means ~0.7 to 0.8 Ah stored. The pack was allowed to rest and the measured:

Cell Bank / Voltage (V)
1 ............ 4.1417
2 ............ 4.1403
3 ............ 4.1414
4 ............ 4.1413
5 ............ 4.1401
6 ............ 4.1146
7 ............ 4.1384
8 ............ 4.1403
9 ............ 4.1428
10 .......... 4.1411

What is interesting is that the maximum difference has been increased by going above 41V charging to 28.2mV. No sign of any balancing occurring. The next step was to charge to 41.6V. A further 0.156Ah was applied. The voltages were:

Cell Bank / Voltage (V)
1 ............ 4.1625
2 ............ 4.1606
3 ............ 4.1617
4 ............ 4.1617
5 ............ 4.1600
6 ............ 4.1290
7 ............ 4.1582
8 ............ 4.1603
9 ............ 4.1634
10 .......... 4.1613

Overall pack voltage 41.581V

Now the maximum difference has increased to 34.4mV. What is also interesting is that some of the cells are higher than 1/10 of the applied charging voltage of 41.592V. Still no sign of any balancing going on. So next step was 41.7V. This should get interesting as this is the voltage the old Yose Power SANS chargers used to be set to. A further 0.067Ah was applied, so energy input is dropping off.

Cell Bank / Voltage (V)
1 ............ 4.1744
2 ............ 4.1725
3 ............ 4.1736
4 ............ 4.1737
5 ............ 4.1720
6 ............ 4.1381
7 ............ 4.1701
8 ............ 4.1716
9 ............ 4.1731
10 .......... 4.1736

Still no sign of balancing, bank 6 was still low, even after several hours. The current draw from the charger was 0.7 mA, so I left it on over night to see if it would balance by morning:

Cell Bank / Voltage (V)
1 ............ 4.1764
2 ............ 4.1763
3 ............ 4.1757
4 ............ 4.1781
5 ............ 4.1777
6 ............ 4.1416
7 ............ 4.1718
8 ............ 4.1685
9 ............ 4.1763
10 .......... 4.1764

The maximum difference was now 36.5 mV, so no balancing. Hence, the SANS power pack that came with the battery at 41.7 V would not have balanced the battery.

I then tried 41.8V. Something seemed to start happening. The power input went up to 0.183Ah. I disconnected after running for several hours and then on the Monday morning I measured the voltages:


Cell Bank / Voltage (V)
1 ............ 4.1738
2 ............ 4.1766
3 ............ 4.1721
4 ............ 4.1763
5 ............ 4.1791
6 ............ 4.1551
7 ............ 4.1693
8 ............ 4.1658
9 ............ 4.1657
10 .......... 4.1740

Overall pack voltage 41.705V

So the maximum difference has decreased to 24mV. It seems as if some balancing has occurred, but not enough. The next step was to go to 41.9 V. Charging was performed at 41.903V, 0.125Ah was applied. I was monitoring the voltages closely and I was getting a bit concerned about bank 10 as this was over 4.2V per cell.

Cell Bank / Voltage (V)
1 ............ 4.1904
2 ............ 4.1933
3 ............ 4.1888
4 ............ 4.1931
5 ............ 4.1964
6 ............ 4.1837
7 ............ 4.1859
8 ............ 4.1820
9 ............ 4.1822
10 .......... 4.2018

Overall pack voltage 41.898V

Finally, bank 6 is no longer the lowest bank, although the voltage difference between bank 6 and bank 10 is still 18 mV.

I left it charging for several hours then switched it off to rest for a couple of days. I was getting worried as this was pushing one bank over what I was happy with.

Cell Bank / Voltage (V)
1 ............ 4.1741
2 ............ 4.1769
3 ............ 4.1721
4 ............ 4.1767
5 ............ 4.1795
6 ............ 4.1601
7 ............ 4.1664
8 ............ 4.1651
9 ............ 4.1656
10 .......... 4.1744

Overall pack voltage 41.711V

So after relaxing for a while, bank 6 is the lowest value again and the voltage difference is 19.1mV, which is about the same (or a little worse) as what I started with after the pack hadn't been balanced for years.

The things to take away from this exercise:

Balancing did not start until over 4.18V per cell.

There is negligible additional capacity above 4.15V/cell and only about 5 to 6% between 4.1V and 4.15V/cell

The balancing wasn't all that effective compared to no balance at all, the cells seemed to drop back to where they were comfortable after charging. It is also a question how accurately each balancing module in the BMS can measure and hence balance voltage.

It's easy for a good bank to be driven over 4.2V when the BMS tries to balance - even if the overall applied voltage is less than 4.2V/cell, it can't bleed away enough current. The bigger the difference in the banks, the bigger the problem this will be. So if charging to 4.2V every time, the good banks are going to get a lot of stress. If the charger is over 4.2V/cell I imagine the voltage on the best banks will approach harmful levels and drastically shorten life.

It seems to me the best way of making sure a pack stays in balance is to make sure the cells are matched properly in the first place. I think pack manufacturers should print the bank voltages, capacities and internal resistances on the pack when it is new to show that it has been constructed properly.
Did you check at what voltage the balancing transistors open? Where did you get 4.18v from? That sounds a bit high. Did you use a calibrated device to check your charge voltage?
 

WheezyRider

Esteemed Pedelecer
Apr 20, 2020
1,690
938
Did you check at what voltage the balancing transistors open? Where did you get 4.18v from? That sounds a bit high. Did you use a calibrated device to check your charge voltage?

It's using a Fluke DVM. The figure of 4.18 V is what I have also seen on the data sheets for other BMS modules, so I was not that surprised.
 

Nealh

Esteemed Pedelecer
Aug 7, 2014
20,916
8,529
61
West Sx RH
Thanks for doing this WR , I know you were doing this early on but wasn't sure you were continuing.
Over all that is pretty damn good for no name china cells.
Balance can vary a little and depends on the specs of the BMS in use.
 

saneagle

Esteemed Pedelecer
Oct 10, 2010
6,791
3,132
Telford
Thanks for doing this WR , I know you were doing this early on but wasn't sure you were continuing.
Over all that is pretty damn good for no name china cells.
Balance can vary a little and depends on the specs of the BMS in use.
Any battery will stay in balance if the balancing is active.
 

WheezyRider

Esteemed Pedelecer
Apr 20, 2020
1,690
938
So you didn't check when the transistors opened?

Not easy to get to the transistors when they are under insulation and heat sinks. However, nothing happens in terms of balancing at 4.17V, even when left overnight, while at 4.18V something starts to happen but it is slow. It is not until 4.19V you see quite rapid change. So you can tell what is happening by looking at the data, even if you can't get to the internals. Yes I could rip it apart and look inside, but I do want to use this BMS.

51692

51693

This is a parameter sheet from the higher end BMS modules I have been buying, showing that balancing does not start until 4.18V

51694
 
Last edited:

WheezyRider

Esteemed Pedelecer
Apr 20, 2020
1,690
938
Any battery will stay in balance if the balancing is active.
Which it is not when charging to 41V, yet if the battery is made properly in the first place with well matching cells it seems to stay in balance quite well, even after many cycles, while the battery being forced to balance every cycle will be putting some banks through a lot of unnecessary stress and shortening the life of the pack.
 
Last edited:

Nealh

Esteemed Pedelecer
Aug 7, 2014
20,916
8,529
61
West Sx RH
Any battery will stay in balance if the balancing is active.
Most BMS are simply passive balancers and only balance at the top end voltage , as you know via the bleed resistiors. But a wider difference in voltage may not balance out in one go
An active balancer works a bit different and will balance at any time during the voltage range if a wider imbalance is detected, these are vary rare on generic ebike batteries.
 
  • Agree
Reactions: WheezyRider

WheezyRider

Esteemed Pedelecer
Apr 20, 2020
1,690
938
Thanks for doing this WR , I know you were doing this early on but wasn't sure you were continuing.
Over all that is pretty damn good for no name china cells.
Balance can vary a little and depends on the specs of the BMS in use.
I've been charging all my packs to 41V for quite a while now. The Yose power packs and the ones I've built myself from single batches of cells have never had a balance problem. I'm currently experimenting with a small scooter pack I've put on on a child's bike, we'll see how that goes after 6 months or so. On the main bike at the moment I'm running a Yose Power Silverfish, it's been through the winter, at least two cycles per day and hammered hard in cold conditions and even after more than 400 cycles it's working as well as day 1, which is the advantage of only charging to 4.1V/cell, the increased longevity.
 

WheezyRider

Esteemed Pedelecer
Apr 20, 2020
1,690
938
The test is pointless without determining whether the BMS is actually balancing or not.
Well, I think you can see that from the data, unless you chose to ignore it.

You've told me you have a garage full of battery packs, why don't you try the experiment and report back?
 
  • Like
Reactions: StuartsProjects

saneagle

Esteemed Pedelecer
Oct 10, 2010
6,791
3,132
Telford
You've told me you have a garage full of battery packs, why don't you try the experiment and report back?
I did, and found that regardless of specification, some do and some don't, which is why I asked you to check which category yours falls into.
 

WheezyRider

Esteemed Pedelecer
Apr 20, 2020
1,690
938
Most BMS are simply passive balancers and only balance at the top end voltage , as you know via the bleed resistiors. But a wider difference in voltage may not balance out in one go
An active balancer works a bit different and will balance at any time during the voltage range if a wider imbalance is detected, these are vary rare on generic ebike batteries.
It's a bit of a worry what you have observed with the Daly BMS modules, sometimes discharging banks. Something like that could be a real issue in terms of balancing. I wonder how big a problem this is with other BMS brands.