SOC vs HV/LV delta and balancing….

[EVentropy]

The following observation I should have done when car was newer. Or better, new, with periodic data points as it aged. Alas, I was occupied with other things.
Car was charged to 100% and after a ~4hr rest:
HV= 349.xx
SOC =100% info 103.xx% car scanner
HV/LV delta = 170mV. Car scanner ( 3.44V, 3.51V)

Drove to store 2km and delta had reduced to 50mV.
Drive up hill a bit 500’ vert delta dropped to 10 mV,
Possibilities?
1. My car is not balancing and is faulty and bms is set to highest cell reaching approx 3.51v to disconnect charge bus.
2.Balancing is somehow done via discharge bus, or surface charge blows off high cell first and this is just characteristic of LiFePO4. And MG’s blurb is bs.
3………
Merry Xmas and the best discharging to you all!

 

[Coulomb]

Drove to store 2km and delta had reduced to 50mV.
Drive up hill a bit 500’ vert delta dropped to 10 mV,
Possibilities?

LFP has a very flat voltage versus SoC curve, except at very high and very low SoC. After driving a bit, it got to a lower SoC, but still not quite on the very flat part. Hence the 50mV delta. After the hill, you were on the very flat part, hence the 10mV delta.

It is just a characteristic of the LFP cell chemistry. So the cell delta information will really only be useful just after a full charge.

 

[EVentropy]

Yep that all makes sense but why doesn’t MG4 “balancing facility” do that automatically when it hits 100%! Four hours at 100% should be plenty of time. Surely the balancing facility doesn’t shut off at a delta of 0.15V. And if so, what was the delta before it commenced balancing? I am leaning towards its is not enabled in software, or it is faulty. The reducing of the the high cell back into flat zone by use of the vehicle is irrelevant.
Unbelievable that there is no manufacturers documentation of this process.

 

[Coulomb]

Four hours at 100% should be plenty of time. Surely the balancing facility doesn’t shut off at a delta of 0.15V.

It depends on how unbalanced the battery is. A newly bought car has probably sat in yards for months, not getting a decent chance to balance.

 

[EVentropy]

The only bms I have encountered so far, and a budget one at that, you configure to when it starts actively balancing, and to what it will shut off at, which is recommended lower than commencement. (Obviously)
I am looking forward to talking with the service technicians and hopefully be reassured by their level of knowledge. As it’s Xmas break I am going to discharge to <5% and do a slow charge and monitor for several hours after achieving 100% with as little powered up as possible, Although I believe the computer is powered from aux battery so that shouldn’t be relevant.
Feel free to add reasons why this is not useful or objective. It will be more than I know now, to go to the techs with.

 

[NLMGSAN]

The following observation I should have done when car was newer. Or better, new, with periodic data points as it aged. Alas, I was occupied with other things.
Car was charged to 100% and after a ~4hr rest:
HV= 349.xx
SOC =100% info 103.xx% car scanner
HV/LV delta = 170mV. Car scanner ( 3.44V, 3.51V)

Drove to store 2km and delta had reduced to 50mV.
Drive up hill a bit 500’ vert delta dropped to 10 mV,
Possibilities?
1. My car is not balancing and is faulty and bms is set to highest cell reaching approx 3.51v to disconnect charge bus.
2.Balancing is somehow done via discharge bus, or surface charge blows off high cell first and this is just characteristic of LiFePO4. And MG’s blurb is bs.
3………
Merry Xmas and the best discharging to you all!

Is this menu only available with a LPF battery? I can't find it in my ER.

 

[Everest]

Drove to store 2km and delta had reduced to 50mV.
Drive up hill a bit 500’ vert delta dropped to 10 mV,
Possibilities?
1. My car is not balancing and is faulty and bms is set to highest cell reaching approx 3.51v to disconnect charge bus.
2.Balancing is somehow done via discharge bus, or surface charge blows off high cell first and this is just characteristic of LiFePO4. And MG’s blurb is bs.
3………

3.... absolutely normal and expected for LFP, as @Coulomb explained above.

Yep that all makes sense but why doesn’t MG4 “balancing facility” do that automatically when it hits 100%! Four hours at 100% should be plenty of time. Surely the balancing facility doesn’t shut off at a delta of 0.15V.

Balancing will typically take place above 3.4V/cell whilst charging. Once one cell hits the max configured voltage (guessing might be 3.5V) charging and hence balancing will stop.

And if so, what was the delta before it commenced balancing? I am leaning towards its is not enabled in software, or it is faulty. The reducing of the the high cell back into flat zone by use of the vehicle is irrelevant.
Unbelievable that there is no manufacturers documentation of this process.

The only issue I see is where you say... "HV/LV delta = 170mV. Car scanner ( 3.44V, 3.51V)"

Your data is wrong. 3.51-3.44 = 70mV, not 170mV.

What matters is how much energy is stored and whether an imbalance has materially reduced you 51kWh and/or range.

 

[EVentropy]

Sorry -typo. 3.34V
170mV is correct.

 

[Everest]

Sorry -typo. 3.34V
170mV is correct.

Ah - OK. Ignore what I wrote above then. 3.34V is too low. As discussed on the thread about LFP SOH and balancing recently, the cell voltages will need to be min of 3.4V before balancing should take place, IME.

We don't know (and will probably never be privy to) info about what type of balancer SAIC use (active or passive), what the max balancing current is, at what rate the LFP's are charged once cells get to the upper knee or, indeed, what max cell voltage they get charged to.

Hence in your situation, I would attempt as many balances as possible and observe both the delta and, more importantly, the min voltage of the lowest cell. Then repeat balancing several times until the min cell reaches at least 3.4V, preferable a bit higher. I'd be content if you can get them to (say) 3.45V and 3.55V with a delta of 100mV (assuming the highest cell is charged up to 3.55V, that is.

In order to do said balancing, it should not be necessary to fully discharge the battery, IMHO. Discharging to around 80 to 90% should be fine - at that point the cells should be fairly close in voltage around 3.30-3.35V, no load voltage (see below).

Then charge back up to 100%, rinse and repeat.

Graph below is no-load voltage vs. SoC for typical LiFePO4 cells

 

[siteguru]

FWIW an earlier version of the iSmart app showed charging power to 2 decimal places. During balancing we'd typically see .08kW (80W) charging power on the LFP pack. I don't know if that snippet of info helps.

 

[Everest]

Yes, that's interesting - that's an incredibly low current (around 0.2A) so less than 0.002C charge rate. That would imply the balancing could take a very long time e.g. 10's of hours, if not days, with small amount of excess voltage bled off the highest voltage cells as it progresses.

Is that the experience of LFP users? i.e. can the balancing stage go on for so long? Is there some indicator that tells you when the pack is being balanced, or does the car just sit at 99% SOC for ages before getting to 100% and charge stopping?

 

[siteguru]

In the app we see that the SoC reaches 100% and the finish time equals the current time - and they both keep incrementing whilst balancing. 30-40 minutes seems typical balancing time for a pack that is fairly well maintained. (From memory - I've not monitored mine for a long time now as the novelty wore off).

 

[EVentropy]

Is this menu only available with a LPF battery? I can't find it in my ER.

I doubt NMC batteries makes any difference. Connect your OBD2.
Plug your phone in and select CarPlay on info centre (either gender) /home /car scanner will come up on info centre with a couple of PIDs from a well known utuber, forget his name…..
A bit underwhelming……

 

[EVentropy]

We don't know (and will probably never be privy to) info about what type of balancer SAIC use (active or passive), what the max balancing current is, at what rate the LFP's are charged once cells get to the upper knee or, indeed, what max cell voltage they get charged to.

Hence in your situation, I would attempt as many balances as possible and observe both the delta and, more importantly, the min voltage of the lowest cell. Then repeat balancing several times until the min cell reaches at least 3.4V, preferable a bit higher. I'd be content if you can get them to (say) 3.45V and 3.55V with a delta of 100mV (assuming the highest cell is charged up to 3.55V, that is.

In order to do said balancing, it should not be necessary to fully discharge the battery, IMHO. Discharging to around 80 to 90% should be fine - at that point the cells should be fairly close in voltage around 3.30-3.35V, no load voltage (see below).

Then charge back up to 100%, rinse and repeat.

Yep, sounds like a plan! Can’t see the point if batteries are within a few mV at 80ish%.
I top balanced my boat batteries by fast charging in series and then paralleled on a lab charger at 3.6V, to a 1-2mV delta.
Then discharged to 0% (4s config) and immediately charged back to a tail current of <0.5A. ( after a bit of auto balancing) Capacity checked out. Then calibrated my SOC to 100% and SOH to 99% (don’t know why I couldn’t get that last 1%) This was TEDIOUS, and as above, novelty is no longer there. Two years later they are working fine, but SOC is not correct and SOH is 95%. BMS’s disconnect before the knee if I ignore the impending audible warnings.

Obviously when the highest cell hits the knee it’s voltage goes exponential quickly, triggering charge termination. There on in, in my case, I “need “ to monitor what going on and see if I can get the low cell(s) up.

Surprisingly enough my bms balancer on the boat bats is only 150mA and seems to do a sterling job. Imbalance dependent of course.(I have Heltec 5A drop ins but never got around to installing.)

 

[EVentropy]

In the app we see that the SoC reaches 100% and the finish time equals the current time - and they both keep incrementing whilst balancing. 30-40 minutes seems typical balancing time for a pack that is fairly well maintained. (From memory - I've not monitored mine for a long time now as the novelty wore off).

Don’t quite follow you……
Soc reaches 100%- finish time equal= ct
Then both keep incrementing -both? Is there a stated time to balance and it increments as to progress?
Does it say balancing finished at end?
I see nothing,
I have never used the app for charging. Guess I will have to run two iPhones, one for app and one for car scanner.

 

[T1 Terry]

I wouldn't hold much faith in any LFP or LYP voltage based graph relating to SOC, the chemistry just doesn't work like that. As far as the 3.34V - 3.51V delta, the 3.51V cell is fully charged if it can hold that voltage for a few hrs at least, the other cells are not fully charged, so the battery is not at 100% SOC by a long stretch.

The pitiful thing about most BMS systems is the poor balancing capability regarding getting the whole battery fully charged.
In general, most cell balancers use a resistor set up that burns tiny bits of capacity off the high cells.

I think the consensus of opinion using known data suggests the cell capacity in the LFP MG4 battery is somewhere around the 125Ah to 130Ah. Most resistor type balancers aren't anything like 0.2A, that would mean around 1W of heating for each cell as they reached fully charged voltage, that's getting up around 100W of heat being generated that needs to be removed without doing damage to the cells that are already getting internally heated. No energy transfer is 100% efficient, the inefficiency is electrical energy changing to heat energy rather than stored energy going in or out of each cell.

So, with that in mind, the resistor would be burning off maybe 20mA and not 200mA, or 0.02amps and not 0.2 amps.

Regardless, even using the 0.2 amps as the balancing current, it would take 5 hrs to burn off 1Ah of stored energy.
As the voltage on the high cell drops, that balancing current will reduce, what is needed is for the charging to resume at a trickle charge rate once the high cell drops below the high cell charge cut voltage for a given length of time, say 15 mins so 0.25Ah has been removed from the high cell, that would have dropped the voltage to closer to the 3.45V that is considered the rested upper voltage of a fully charged LFP cell without it pushing the voltage higher ......

This will of course push the high cell back up to the cell charge cut voltage, but it would have put 0.25Ah into all the other cells, allowing them to catch up ..... as you can see, even with this technique it would take hrs to balance a battery.

After market produce induction cell balancing, the capacity from the high cell gets transferred to the lower cells, much faster and a greater current can be shifted from one cell to the others because far less heat is involved ...... but probably too expensive for manufacturers to build into their systems, so a long slow rebalance is the only solution.

Someone needs to design and build a circuit that turns off the charging power when the cell cut voltage stops the charging, then turn it back on say 15 mins later and repeat until the target battery voltage is reached, then the 100% SOC can be believed to be accurate, with any cell below 3.5V rested, the battery is not at 100% SOC.

T1 Terry

 

[siteguru]

Don’t quite follow you……
Soc reaches 100%- finish time equal= ct
Then both keep incrementing -both? Is there a stated time to balance and it increments as to progress?
Does it say balancing finished at end?
I see nothing,
I have never used the app for charging. Guess I will have to run two iPhones, one for app and one for car scanner.

I meant exactly what I said ... whilst balancing is underway the SoC in the app shows as 100%, but still shows as Charging, and the Finish time keeps incrementing as per the current time. Once complete the car tells my charge point to turn off - i.e. go to standby/idle mode - and the app says Not Charging. (I'm never in the car so I've no idea what the car says - the app gives no further info than what I've said).

 

[EVentropy]

Excellent posts T1 , and Everest, food for thought.

I think my balancer is either not working or is a primitive resistive one as I roughly noted delta vs soc at intervals from 98% through 100% until it started dropping.

Data points of interest, without spreadsheeting it.
At 100% SOC:
Charging terminated, max cell V…..3.66V*1
Min cell V ……………………………3.37V
Delta………………………………….290mV
Pack V = 351V, mean =360/104..…3.375V
Note mean is very close to min cell voltage so
*1 is an outlier that has jumped the knee and gone expo?

On supposed “balance mode”, after 1.5hrs, zero load: (kW meter)
SOC…………………99.8%
Max cell V……………3.44V
Min cell V……………..3.34V
So obviously no topping from high cells or ext charger.
I am aware of the LiFe 3.65v hv bms charge bus disconnect as a spec from all cell manufacturers but I was a little surprised it was set that high and the “fuzzy logic” on the 100% SOC. Like HV =1; goto 100%.

A couple of questions for the afflicted:
1.Anybody know how fast surface charge dissipates without intervention? ie 3.xx to 3.4ish surface charge to full, rested.
2. What is your Pack voltage MG4 51kWhr at SoC ~100% after “balancing”, rested.

 

[Coulomb]

Pack V = 351V, mean =360/104..…3.375V
Note mean is very close to min cell voltage so
*1 is an outlier that has jumped the knee and gone expo?

Yes, it sounds like one cell is considerably higher voltage than the others. So your battery is unbalanced. I would keep attempting to charge and balance it with say an hour settling time in between to try and get the average cell voltage closer to 3.50 Volts Per Cell after a charge / balance / rest for an hour or two.

 

[T1 Terry]

A couple of questions for the afflicted:
1.Anybody know how fast surface charge dissipates without intervention? ie 3.xx to 3.4ish surface charge to full, rested.
2. What is your Pack voltage MG4 51kWhr at SoC ~100% after “balancing”, rested.

1) I can answer the first one:
A) if the cell is not saturation charged, it will be absorbed into the cell within an hour or so.

B) if the cell is saturation charged, it will remain higher than 3.4 to 3.5V for a month or so, then cell degradation will have occurred and the excess will be burnt off due to higher internal resistance and chemical breakdown in the electrolyte.

2) 3.5V x 104 cells = 364 VDC .... this is dependant on full balancing being achieved, if the OEM BMS can achieve this is yet to be seen ........

T1 Terry