LFP Car Battery Not Balancing - Cause for Worry?

[wattmatters]

Sometimes balancing is noticeable: The charging stays on 99% for a long time. I had this on my previous (non LFP) EV. But haven't seen such a behavior on my (R.I.P) BYD EV w/LFP.

In my car the car's display will tell you it's balancing, but in general you are never in the car when it does that.

 

[sunsurfer]

Balancing isn't a big deal. My car has a 104 cell series pack, and it balances at end of reaching its set charge limit, normally finished balancing in 5-10 minutes. If you do a full charge then it might take longer, how long it takes depends on how long since the last full charge and how much DC fast charging you do. e.g. just a little while ago mine finished its charge to a set 80% SOC. It balanced for 8 minutes at the end:

The change in charging power is due to automations that:
i. ensure I don't over draw the current limit on that phase during the free energy period (12-2 PM)
ii. match excess available solar PV

Right at the end can see the period when the car is performing a cell equalisation, power draw drops to ~100+/-50 W, most of that is just keeping the car's system on/operating during this process and trickle charging the 12 V battery. Once done, power draw is stopped by the car.

I suspect most good OEM packs have no issues with premium and well matched cells. Its the "solar grade" packs that will be coming in the future but I guess we will have to see in a few more years.

 

[wattmatters]

I suspect most good OEM packs have no issues with premium and well matched cells. Its the "solar grade" packs that will be coming in the future but I guess we will have to see in a few more years.

The battery pack will outlive the car so I'm not overly concerned.

 

[meetyg]

The battery pack will outlive the car so I'm not overly concerned.

Yeah, I think it's more about getting the full capacity of the battery, more range.

 

[PeterWA]

Can note also that drive battery charging continued for ~ 20-minutes after SOC had reached 100.0%. That's because the SOC readings drift/lose accuracy over time.

Same for me.

I found my old backup files (had kept the data in Home Assistant, but the database got too big and I had to cull it). Looked at every single time I charged to 100% since I got the car in November 2022. Always the same. Charging. At some point SOC changes to 100%. It continues a little at 9A (anywhere between 0 and 30 minutes), then drops to 3A for 3 minutes, then straight to 0.

I never saw anything that would indicate balancing. I don't think it ever balanced, unless the battery can balance completely internally, without any AC current supplied through the charger. In theory that would be possible, the BMS will have power from the HV battery and can burn off charge from the highest cells through resistors. But then I never saw an improvement in cell imbalance. So odds are my HV battery never did a balance...

This is why MG also recommend occasionally performing a full change following a deep discharge (to <= ~10%) as it recalibrates the BMS SOC and SOH readings.

I'm trying that now. Not driving much this week, it will take a couple more days to get below 10%.

Oh, btw: not sure if it needs to be <10%, the only time I saw a jump in SOH was after a charge from ~20% to 100%. Which also was the only time I did such a charge. I have been around 20% several times, but always followed with a charge to betwen 80 and 90%, except that once. The other times I got below 20% was on long trips with DC charge.

 

[wattmatters]

Oh, btw: not sure if it needs to be <10%

Relevant pages from the user manual. Note the comment on p. 293 which I take to mean charging from less than 10% charge state.

 

[CaliSunHarvester]

Relevant pages from the user manual. Note the comment on p. 293 which I take to mean charging from less than 10% charge state.

View attachment 281383
View attachment 281384
View attachment 281385
View attachment 281386
View attachment 281387
View attachment 281388

Very interesting. Just to confirm: this applies to an EV with a LiFePO4 battery, right?

I need to look at my EV's manuals. They are still sealed.

 

[meetyg]

Very interesting. Just to confirm: this applies to an EV with a LiFePO4 battery, right?

I need to look at my EV's manuals. They are still sealed.

Yes, for EVs with Lifepo4.

 

[wattmatters]

Very interesting. Just to confirm: this applies to an EV with a LiFePO4 battery, right?

The manual covers all their MG4 models in Australia (other than the XPower) which include LFP and NMC. They have three pack types here, 51 kWh LFP, 64 kWh NMC & 77 kWh NMC. But the XPower manual on these sections will be the same as it has the same 64 kWh NMC battery.

 

[roly1]

My experience with my mg4.

What to ask the HV tech if your car plotted this.......

If you not into how 100+ cells are managed, don't read as somewhat OCD. I discovered this some months ago but the car performed fine and I was too busy at work to pursue. I recreated as I didn't save and I have booked my first service, this is as good a time as any to ask the experts. Last...

www.mgevs.com

My thread on MGEV

This is not how my 4yro LiFePO4 CALB cells
Balance in my boat. Far from it.
I’m going to do 1 more plot as above.
1. From regular recharge base voltage ~25% , plot from 98% and leave charger on for 10hr after 100%
2. From zero or close and do same.
Then I’ve covered pretty much all I can think off.
Charging , balancing,then blowing HV cell to allow bms to start charging does not improve low voltage cell or mean pack voltage. As l have tested that.

 

[roly1]

Hi Peter, You are on the same tack as me wrt to MG4 LiFePO4 balancing. Sorry for dumping above in this thread without any preamble as I was occupied doing other stuff.
I have many questions for the HV tech (service) this thursday as what is suggested in the manual, and by other users
doesn't stack up in practice. eg delta after 15hr still plugged in is still ~100mV!
Your plotting, at a glance, concurs with mine.
As stated, ad nauseum, my boat BMS/balancer is graphically transparent, and active as opposed passive so I guess just
to accept that MG have built the monitoring system of the MG4 to a minimum efficiency/redundancy/safety as the chemistry
is so forgiving. (Relative to others) If anyone has any data on the excite model it would be very much appreciated, and/or tack
to establish what is expected of the BMS's ex-factory.

My boat is 8x CaLB 230Ahr 4S2P, 2x BMS's 200A 150mA balancing HV disconnect set at 3.5V . (Needs more, for optimum, have 5A Heltech)
4yro. I know "100%" is somewhat holy grail territory, but my boat can get pack voltage to 3.5V at 0.02C tail current without do much phaffing around. I know ITO capacity there is little point in being in the "knee" , but with a LV cell of < 3.4V is sub optimal IMO. (For supposedly matched cells)

Any critique of my conclusions welcome.

 

[roly1]

Yes, OBD2. Via an open source piece of hardware called OVMS that stays connected permanently. It feeds data into Home Assistant.

The battery cell data makes sense with typical charging behavior. Around 2mV cell deviation in the flat part of the curve (anything below 3.3V). Deviation growing rapidly when the battery approaches 100%.

I wouldn't care if that was happening when all cells are over 3.45V, or if some balancing was apparent. But seeing this happen when most cells are still below 3.4V and having the charge current drop to zero is worth looking into more closely.

My sentiments exactly! See my plot. I even did a discharge to 98% and then quick AC recharge and balance results were worse. Low cell didn't even get to 3.4V! Bizarre drop in low cell on balancing hi cell that is not consistent with surface charge decay of low cell. (Like it is also being resistively reduced) Could be some inaccuracies in the app, monitoring devices~ fuzzy logic etc

Then the only hope for an improvement will be the parameters of BMS updated via SW for longer burn off off high cell, continued charging of
low cell(s) rinse & repeat until more satisfactory mean pack voltage is achieved, as you suggested.
Anybody like to speculate how consultation with MG tech is going to go?

 

[455dan]

@Brucey guessed right, it's an MG. CATL would have been a give-away. BYD make everything in-house with their blade batteries.

Nevertheless thanks so much for the detailed reply. Looks like there are many more similarities under the hood between those brands than I expected.

Good to know that you have seen similar cell voltage deltas. Maybe there is nothing to worry about.

From my home battery I know that 3.35V is in the bend in the LFP voltage curve. I charge to 3.45V (bulk, then absorption), then drop to 3.35V for float. That drop drains 2% of the battery capacity, but that float voltage is reached when the battery is essentially resting. The car has most cells at 3.37 to 3.38V when charging just stops, not at rest. I'll have to have a closer look where the cell voltages settle to after that. Best estimate right now: around 3.33V for the bulk of the cells. That would be at least 5% below full capacity, perhaps significantly more than 5%.

One thing we definitely can agree on: the balancing circuitry (and programmed logic) is very much lacking. I can't understand how car manufacturers can be so careless, seeing what long warranties they give. MG now 10 years.

I have a trip coming up in a month that last year used up 90% of the battery by the time I get to the only charger on the way. I hope I'll still make it this year. The scenario 10% down to 0 in an instant is exactly what I fear might happen with a large imbalance of the HV battery.

Not related to the battery but as far as MG/ Roewe/SAIC

I really wanted the 3 motor Marvel X back when they first showcased it.

Still prefer the front interior design to my model Y. The center wheel display with Augmented Reality Power moonroof and sunshade were cool. They were also the first to test out Inductive car charging.