Replace or Repair my worn 48V Ebike Battery? How do I extend the life from 450 cycles?

[raming]

My $300 Asian Ebike has an 7.8Ahr 48V Battery pack that failed at 444 Charge/Discharge Cycles. Here is my experience replacing and repairing the Battery pack.

Use Details:

26" Wheels, 21 speed, 75lb steel frame, 750W Peak Motor with pedal assist and/or throttle.

Battery: 13Sx3cell parallel pack consisting of 2.6Ahr 4.2V 18650 matched cells. Integrated BMS with Passive Balancing Circuit.

8 mile 35minute bike ride on hilly terrain consuming about 3Ahr.

Charge several hours from 'red' to 'green' balanced condition after every use.

Full Charge Voltage: 54.8V (4.215V/Cell). Balancing starts in 'green mode' at 4.1V/Cell.

I replaced the failed pack with a larger 10Ahr 48V battery but realized that all the major 18650 LiPO cells all have a rated cycle life of only 300 to 500 cycles! This means that at one 3Ahr 35minute ride per day followed by a full ‘Balance Charge’; one needs to replace the ebike battery pack every 1.5 to 2 years. Therefore, I decided to repair the old ‘worn’ battery pack and see if I could extend the life well beyond 500 cycles.

Key take-aways: Battery balancing may not be necessary and Full Discharge (below 3.6V/Cell) and Full Charging (greater than 4.05V/Cell) limits battery cycle life. (By using the battery between 20% and 80% capacity, there are reports that cycle life can be extended to up to 5000 cycles).

• Battery pack failed because one of the cells in the 13Sx3 pack discharged to 3.0V while all the other cells were closer to 4.0V. (The BMS shuts down the pack output when any cell drops to 3.0V or less). The ‘Low V’ Cell has similar 90% SOH (State of Health) as a ‘Good Neighbor’. However, it lost capacity 28 times faster (14% Voltage loss versus 0.5% over 30 days at room temperature). This abnormal leakage could not be resolved with the internal passive balancing circuit on the BMS. (See attached graph.)
• 3 more cells failed after the initial repair after only 24 more use cycles with the same low volage failure signature.
• The Battery pack was modified with a 14 pin external port so that all 13 cell groups could be monitored and custom balanced. (See attached photos). Since the 12 new 2.6Ahr non-exact replacement cells could not matched to the older cells; all the cells were manually balanced to 3.65V +/- 0.010V. (See attached graph). The cell miss-matching does not cause problems since none of the cells are stressed outside of the 20-80% safe operating envelope.
• The Repaired battery pack was successfully used for 25 cycles where pack was used only between 20% and 80% charge level using a non-balancing timed charge.
• Battery balance was preserved at 3.65V after 25 cycles and all 13 cell sets had low leakage (low self-discharge).

Bottom Line: The initial results are encouraging. I may have salvaged my aged pack by cycling the battery only between 20% and 80% of capacity and foregoing and full charge/balancing. I am no longer fully charging my new 10Ahr battery pack and my 3Ahr typical discharge is easily achieved within the 20% to 80% capacity range. Note: A timed 1hour charge with the standard charger is sufficient to charge the battery after the typical 3Ahr ride.

 

[amberwolf]

FWIW, balancing doesnt' fix anything. All it does is make all the cells the same voltage at some point in the charge curve (usually full). For more details on that, you can look up my other posts with that same statement.

Essentially, if you want to completely "fix" a battery with unmatched cells, and not worry about balancing, you'd need to replace all the cells with matched ones (identical (not just similar) properties, internal resistance, capacity, etc), so that they all always stay at the same relative voltages because they are all at the same actual SoC because they all stay "balanced" on their own.


There are also a number of threads (and individual posts) discussing, some with good test data, how much lifespan is affected by limiting SoC range, if you want more info on that beyond your own test.

 

[docw009]

Some detailed work there. Nice.

Your larger battery would have lasted a longer time than you expect, if you had continued the previous routine of charging to max and recharging after every ride. 3AH is less than half of its capacity, You were probably down to about 38 volts at the end of the ride, where the smaller pack would likely have been under 36V. Do those estimates come close to what you saw?

How are you getting that 3AH number. I use cheapo wattmeters from the RC hobby sector. Measure the drain during a ride. Measure the power going back in during a recharge from LVC to assess a battery,

"3 more cells failed after the initial repair after only 24 more use cycles with the same low voltage failure signature."

This is is similar to my experience. I had an unbalanced 10S-5P. One group stuck at 3.6 with the rest at 4.1V, In my case, putting in a balance BMS equalized the battery, but the bad group still limited the pack to 60% of original capacity. I replaced all five cells and called it a "win", until another bank failed later. Both groups had a leaker. It wanted to discharge itself and its partners to zero in between recharges, and given time did that. I concluded there's no point in repairing a battery if you have a leaker, As soon as you find one, you;re past the mean time before fail, MTBF, for that lot of cells. I would expect that if your repaired battery is not given constant use, even with its relaxed operating conditions, some of the other older cells will become leakers and eventually pull their groups down without regular recharging,

I am surpised that Vivi puts a balance BMS into their batteries. I used to think that was automatic, but had never seen one in my batteries. I believ that quality cells tend to stay equalized with a nonbalance bms, if run at low currents relative to their advertised C-rates.

 

[raming]

Some detailed work there. Nice.

Your larger battery would have lasted a longer time than you expect, if you had continued the previous routine of charging to max and recharging after every ride. 3AH is less than half of its capacity, You were probably down to about 38 volts at the end of the ride, where the smaller pack would likely have been under 36V. Do those estimates come close to what you saw?

How are you getting that 3AH number. I use cheapo wattmeters from the RC hobby sector. Measure the drain during a ride. Measure the power going back in during a recharge from LVC to assess a battery,

"3 more cells failed after the initial repair after only 24 more use cycles with the same low voltage failure signature."

This is is similar to my experience. I had an unbalanced 10S-5P. One group stuck at 3.6 with the rest at 4.1V, In my case, putting in a balance BMS equalized the battery, but the bad group still limited the pack to 60% of original capacity. I replaced all five cells and called it a "win", until another bank failed later. Both groups had a leaker. It wanted to discharge itself and its partners to zero in between recharges, and given time did that. I concluded there's no point in repairing a battery if you have a leaker, As soon as you find one, you;re past the mean time before fail, MTBF, for that lot of cells. I would expect that if your repaired battery is not given constant use, even with its relaxed operating conditions, some of the other older cells will become leakers and eventually pull their groups down without regular recharging,

I am surpised that Vivi puts a balance BMS into their batteries. I used to think that was automatic, but had never seen one in my batteries. I believ that quality cells tend to stay equalized with a nonbalance bms, if run at low currents relative to their advertised C-rates.

Thanks for the feedback, very helpful!

I used an Eflite PowerMeter connected in series with the battery pack for Discharge Statistics and a Suraielec Watt Meter for charge monitoring. Note that my packs are 13S versus your 10S battery.
Both the repaired 7.8 and the new10 Ahr batteries had very similar start/stop voltage signatures over the same 3.0Ahr 8.3mile bike course:
Battery Vstart Vend Vmin (uphill lowest voltage)
10Ahr 54.5 49.37 46.4 (13s cell voltages of 4.19, 3.80, 3.57)
7.8Ahr 54.62 49.36 46.2 (13s cell voltages of 4.20, 3.79, 3.55)

My initial data about leakers due to wearout matches your experience. However, once I replaced the 4th leaker, I was able, to my surprise, to use the repaired pack for 25 3.0Ahr rides without any more leakers! Note that I charged the battery to only about 4.0V/Cell during this period. Several 2week room temp storage tests confirmed that none of the 13 cell groups had more than a few mV of voltage loss. I read that charging above 4.0V/cell to the full 4.2V/cell will accelerate cracking of the critical Solid Electrolyte Interphase (SEI) layer which ultimately causes battery failure (and leakage?). So, perhaps one can limit the 'worn cell' degradation further by avoiding full charging. I will continue using the repaired battery pack to at least 100 cycles within the 20/80 Sweet Zone to confirm that the pack repair is worthwhile.

Finally, I confirmed that the VIVI BMS has the passive resistor Balancing circuits which apparently are active when the battery approaches full charge (Green Light Mode). My monitoring of each of the 13 cell groups also show that good cells maintain close balance after repeated use cycles and that the balancing circuit is not needed and won't improve significant imbalance due to their very low (<10mA) balance rates.

 

[amberwolf]

I believ that quality cells tend to stay equalized with a nonbalance bms, if run at low currents relative to their advertised C-rates.

It isn't so much whether they are "quality", but rather that they are matched in characteristics, and stay that way.

If they are not matched, they won't stay in balance regardless of cell quality, because they will each charge and discharge at different rates, and voltage drop will be different among them because of different internal resistances.

If they are matched in all characteristics (resistance, capacity, etc), they will all charge and discharge at the same rate, voltage drops will be the same, and they will stay balanced, for as long as all those characteristics are matched.

While quality cells are more likely to start in and remain in this state for longer, simply being quality doesn't say anything about all the cells in a pack being matched to each other in and of itself--that has to be done by the pack builder, if the manufacturer doesn't already do this with the cases / pallets / lots that they ship out--not just guaranteeing a range of values, but actually verifying that they *are* a *specific* value that is identical among them, to as many significant figures as is necessary for the specification and degree of matching.

 

[raming]

It isn't so much whether they are "quality", but rather that they are matched in characteristics, and stay that way.

If they are not matched, they won't stay in balance regardless of cell quality, because they will each charge and discharge at different rates, and voltage drop will be different among them because of different internal resistances.

If they are matched in all characteristics (resistance, capacity, etc), they will all charge and discharge at the same rate, voltage drops will be the same, and they will stay balanced, for as long as all those characteristics are matched.

While quality cells are more likely to start in and remain in this state for longer, simply being quality doesn't say anything about all the cells in a pack being matched to each other in and of itself--that has to be done by the pack builder, if the manufacturer doesn't already do this with the cases / pallets / lots that they ship out--not just guaranteeing a range of values, but actually verifying that they *are* a *specific* value that is identical among them, to as many significant figures as is necessary for the specification and degree of matching.

So, by your analysis, repairing a battery pack with bad cells is not worthwhile because:
1. Worn Cells, by definition, don't match their new twin replacements.
2. Finding the exact same brand of 18650 cells still won't produce a precision match due to batch-to-batch differences.
3. To avoid overcharging and overdischarging, poorly matched cells will have much lower effective capacity.

My points, based upon data are a bit more subtle:
1. I can repair an 18650 LiPO battery pack with similar capacity cells and achieve matching to less than +/-100mV over the discharge curve and still use said pack for 30-60% of its rated capacity. (I elected to actively balance the repaired pack to 3.65V/Cell as seen in the original post graph since this ensures that none of the cell's cause a BMS low-V shutdown near the bottom of the discharge curve).
2. 'Leakage-type Wearout' may be accelerated by charging the cells over 4.0V. SEI cracking seems to be a dominant cause for cell failures and that this happens faster outside the 20-80% 'long cycle life' zone.

Bottom Line: Some of us Ebike enthusiasts really want practical advice on extending the life and reducing the operating costs of the Battery packs. I appreciate data-based feedback as always!

 

[amberwolf]

So, by your analysis, repairing a battery pack with bad cells is not worthwhile because:
1. Worn Cells, by definition, don't match their new twin replacements.
2. Finding the exact same brand of 18650 cells still won't produce a precision match due to batch-to-batch differences.
3. To avoid overcharging and overdischarging, poorly matched cells will have much lower effective capacity.

"Worthwhile" depends entirely on your needs, desires, budget, and intended results.

If you just want to repair a pack, and make it work, then you do whatever it is that makes it work for your purposes.

If you want to repair a pack and have a specific result required, you have to do whatever it is that makes that happen.

If you want to build new packs that work and keep working and stay balanced, then that's a different case.

Same if you want to repair a pack so it "works like new" like a pack that was made that way...it has to be *remade* the same way.

Etc.

Bottom Line: Some of us Ebike enthusiasts really want practical advice on extending the life and reducing the operating costs of the Battery packs. I appreciate data-based feedback as always!

Certainly--I was only responding to the specific point I quoted that the other poster stated.