MG4 51kWh LFP Battery Health

[wattmatters]

By fully recharging after this deep discharge, the electric meter recorded 59.2 kWh, and my terminal indicated 55.05 kWh injected into the battery at 32 A.

Presumably these are AC power supply numbers, not the DC energy going into the battery after conversion losses (which will be significant, of the order of 10%).

 

[Bam Bam]

Presumably these are AC power supply numbers, not the DC energy going into the battery after conversion losses (which will be significant, of the order of 10%).

Good question, does 'terminal' here refer to the ODB scanner report or something else?

 

[king31]

Presumably these are AC power supply numbers, not the DC energy going into the battery after conversion losses (which will be significant, of the order of 10%).

By discharging the battery to a voltage of 2.88 V for the lowest cell and 2.99 V for the highest cell, I was able to charge a total of 59.2 kWh via my 32 A charger. Taking into account the losses related to charging (charger efficiency), the net capacity injected into the battery is 55.05 kWh.



To obtain this precise data, I installed a meter before the charger, which allowed me to precisely measure the amount of energy consumed during charging. This confirms a very interesting gross capacity.

 

[wattmatters]

To obtain this precise data, I installed a meter before the charger, which allowed me to precisely measure the amount of energy consumed during charging.

Can you clarify exactly where these meter(s) are?

I just did a quick experiment to record both the AC power supply to my EVSE as reported by my Iotawatt energy meter and the HV battery DC power as reported by the car via OBD port.

I do need to validate the timestamps, I think there is a bit of an offset but as a check of whether this is valid for data capture it's good enough.

I was fiddling with the CarScanner tool hence less DC data. When I get the chance I'll do some longer duration data capture.

For this little test I was sitting in the car and as a result the aircon was going (it's a hot day, even inside the garage), plus the car would also be charging the auxiliary 12 V battery plus whatever other systems are consuming power.

So a rough loss of 17% from AC power supply to HVDC power into the battery in the above quick test.

As I said, this was not about a test of charge efficiency, just a test of data capture.

 

[Everest]

I was able to charge a total of 59.2 kWh via my 32 A charger. Taking into account the losses related to charging (charger efficiency), the net capacity injected into the battery is 55.05 kWh.

All good stuff, but how do you arrive at such a specific capacity into the battery of 55.05kWh? Are you saying it is the 59.2kWh less estimated 7.5% efficiency?

I would expect more like 10% charging loss from the car's on-board charger, plus a fixed loss from quiescent usage of cars electronics (maybe around 200W if similar to other EV's) over the charge time. That gives a result more like (59.2kWh x 0.9) - (200W x 10.5h) = 53.3 kWh - 2.1kWh = 51.2kWh.

 

[Romeo77]

Wow, a lot of effort there. Bravo!
Always suspected 3.2V nominal was a generic figure for average voltage.
A bit of "fudging" or fuzzy logic going on to keep the battery out of trouble zone and user relatively
happy with performance.
Then there is what Terry said about nominal voltage being under "X" load, with a volt drop?
BYD LiFePO4 have different specs. A little higher.
I am hoping someone can corroborate my experience of 99% (100%) SOC being the triggered when the highest cell reaches 3.65V, and full power charging ceases/balancing starts.
Or is my bank an outlier?

Actualy calculating amp hours with nominal voltage to get kWh just not right because voltage is not constant it changes.
almost all time during charging LFP battery stays at same but not nominal voltage
in my case of BYD with 126 cells in series and 11 kW OBC i see in car scanner ~420 v with 23,7 amps during charging
and when discharged 137 ah down to 9,4% i charged back with that ~10,053 kW for 5 hours and 48 minutes.
that gives me 58 kWh.

 

[EVentropy]

That why I called it an arbitrary benchmarker and is not representative of average voltage either; for my setup NV is at <20%SOC resting. Given that I, V and load are dynamic, SOH can only be definitive with repeat capacity testing. And it can be "massaged" too as people have reported.
I am starting to suspect A grade LiFePO4 cells inherent characteristics lend them selves to a basic cheap balancing system. Cutoff at max cell voltage and switch to trickle whilst discharging high cell to an acceptable V.
Driving car will then finish balancing the bank.

 

[Peter WA]

Interesting thread. Maybe I can add some info:

I personally mostly have experience with a ZS EV (Mk2 currently), but it is very likely that the 51Ah LFP battery is quite similar to the one in the MG4.

My ZS EV is monitored via OVMS. That allows me to track SOH on a daily basis. After almost 2 1/2 years I can say with 100% certainty: the SOH is programmed as a count-down timer that will reduce by 0.01% every single day.

Only once during that time did I see the SOH rise by a little, about 0.5%, in early 2023.

It is likely that that was the only time SOH was actually measured. I have not been able to figure out what triggered that difference to 'business as usual'.

If the MG4 BMS uses the same programmed logic, the behaviour described above is absolutely possible. That SOH can jump by 3% as a result of an actual measurement, that is used to correct the pre-programmed decline by countdown.

 

[Romeo77]

Interesting thread. Maybe I can add some info:

I personally mostly have experience with a ZS EV (Mk2 currently), but it is very likely that the 51Ah LFP battery is quite similar to the one in the MG4.

My ZS EV is monitored via OVMS. That allows me to track SOH on a daily basis. After almost 2 1/2 years I can say with 100% certainty: the SOH is programmed as a count-down timer that will reduce by 0.01% every single day.

Only once during that time did I see the SOH rise by a little, about 0.5%, in early 2023.

It is likely that that was the only time SOH was actually measured. I have not been able to figure out what triggered that difference to 'business as usual'.

If the MG4 BMS uses the same programmed logic, the behaviour described above is absolutely possible. That SOH can jump by 3% as a result of an actual measurement, that is used to correct the pre-programmed decline by countdown.

You probably see calendar degradation that occurs to LFP battery regardless of usage.

 

[Peter WA]

You probably see calendar degradation that occurs to LFP battery regardless of usage.

Maybe that's what MG try to simulate by programming a countdown timer.

But trust me: the car can not measure SOH accurately enough to achieve a linear decline every single day the car is turned on. See chart below. Completely linear (with bigger drops only when the car wasn't turned on for a while or not connected to the WiFi).

What is interesting: the countdown timer was calibrated to stay above the usual 70% warranty threshold for the duration of the warranty. That was back when the warranty was only 7 or 8 years. Now that they increased warranty to 10 years, they might be causing themselves some hassle...

 

[fnegroni]

Maybe that's what MG try to simulate by programming a countdown timer.

But trust me: the car can not measure SOH accurately enough to achieve a linear decline every single day the car is turned on. See chart below. Completely linear (with bigger drops only when the car wasn't turned on for a while or not connected to the WiFi).

What is interesting: the countdown timer was calibrated to stay above the usual 70% warranty threshold for the duration of the warranty. That was back when the warranty was only 7 or 8 years. Now that they increased warranty to 10 years, they might be causing themselves some hassle...

View attachment 35062

Mine reduces by 0.01 too, but after a couple of 5% to 100% charge cycles back to back, it increases by 1-2% and then declines again from there. It seems yours plummets after every bump

 

[Peter WA]

Mine reduces by 0.01 too, but after a couple of 5% to 100% charge cycles back to back, it increases by 1-2% and then declines again from there. It seems yours plummets after every bump

That's interesting info. Thanks!

I don't often get below 20% before I recharge. Maybe that's what is needed. I'll give it a try.

There's just the one single bump up in March 2023. The 'flat' bits are when no data was collected. Away from home, or didn't use the car for a few days. As soon as data was collected again, it dropped straight down to where the straight line decline took it. The internal countdown timer remained ticking, even when data was not read out.

 

[wattmatters]

I don't often get below 20% before I recharge. Maybe that's what is needed. I'll give it a try.

SOH calibration requires a deep discharge followed by a full charge. Only way to reliably measure SOH is testing the full capacity of a battery - all but hitting the minimum and maximum voltages. There are proxies (e.g. pack impedance) but unless and until one does a full charge / discharge / charge then SOH will be a guesstimate.

 

[Tig170]

Maybe that's what MG try to simulate by programming a countdown timer.

But trust me: the car can not measure SOH accurately enough to achieve a linear decline every single day the car is turned on. See chart below. Completely linear (with bigger drops only when the car wasn't turned on for a while or not connected to the WiFi).

What is interesting: the countdown timer was calibrated to stay above the usual 70% warranty threshold for the duration of the warranty. That was back when the warranty was only 7 or 8 years. Now that they increased warranty to 10 years, they might be causing themselves some hassle...

View attachment 35062

Thats what my SE (UK based) has also recorded (via home assist / car scanner), I'm 50k km into this and have not found any major challenges to date, however the system is currently set up as a timer, rather than usage / distance, if one did ask me. (at servicing recently SAIC said it was 100% SOH)

 

[T1 Terry]

You probably see calendar degradation that occurs to LFP battery regardless of usage.

Yet another piece of nonsense put out by the lead acid battery manufacturers and supported by the fossil fuel lobby.
I still have customers that are using their LFP house battery in their caravan, 24/7 for over 12 yrs. Add to that, they are the field kitchen for their son's harvesting business so quite a bit of that time is heavy loads in and out of the battery pack. At the 12 yr service (every 3 yrs for batteries that work hard) they have returned greater than 100% of the advertised capacity using the manufacturers 0.5CA rating, in this case, 200 amps for 2 hrs, an advertised capacity of 400Ah.

Before we lost all the records in the house and motorhome fire ... had a back up copy in each, we had installed over 200 systems, the only ones that didn't last were system that had been in the RV when it was sold and the new owner didn't bother to read the instructions ..... and basically murdered them ..... I still have cells I've recovered from murdered battery packs that are from the very first lot of cells I bought .... these were 3 yrs old in 2011 .... yet they still work fine as jump start batteries and electric winch batteries for skull dragging cars and shipping container onto my truck, so they are 16 yrs old and been treated poorly, yet they still perform every time they are asked to do so .....

Quality LFP cells do not degrade over time ......

T1 Terry

Maybe that's what MG try to simulate by programming a countdown timer.

But trust me: the car can not measure SOH accurately enough to achieve a linear decline every single day the car is turned on. See chart below. Completely linear (with bigger drops only when the car wasn't turned on for a while or not connected to the WiFi).

What is interesting: the countdown timer was calibrated to stay above the usual 70% warranty threshold for the duration of the warranty. That was back when the warranty was only 7 or 8 years. Now that they increased warranty to 10 years, they might be causing themselves some hassle...

View attachment 35062

Are you sure these are LFP cells and not NMC cells? That is more like an NMC cell degradation plot, or it is adjusting the real capacity

T1 Terry

 

[Rolfe]

Yet another piece of nonsense put out by the lead acid battery manufacturers and supported by the fossil fuel lobby.
I still have customers that are using their LFP house battery in their caravan, 24/7 for over 12 yrs. Add to that, they are the field kitchen for their son's harvesting business so quite a bit of that time is heavy loads in and out of the battery pack. At the 12 yr service (every 3 yrs for batteries that work hard) they have returned greater than 100% of the advertised capacity using the manufacturers 0.5CA rating, in this case, 200 amps for 2 hrs, an advertised capacity of 400Ah.

Before we lost all the records in the house and motorhome fire ... had a back up copy in each, we had installed over 200 systems, the only ones that didn't last were system that had been in the RV when it was sold and the new owner didn't bother to read the instructions ..... and basically murdered them ..... I still have cells I've recovered from murdered battery packs that are from the very first lot of cells I bought .... these were 3 yrs old in 2011 .... yet they still work fine as jump start batteries and electric winch batteries for skull dragging cars and shipping container onto my truck, so they are 16 yrs old and been treated poorly, yet they still perform every time they are asked to do so .....

Quality LFP cells do not degrade over time ......

T1 Terry

What did they do to murder them?

 

[wattmatters]

Quality LFP cells do not degrade over time ......

Everything degrades over time, it's just a matter of the rate.

 

[decrep]

Everything degrades over time, it's just a matter of the rate.

the theory is,that even black holes will eventually evaporate.

 

[T1 Terry]

What did they do to murder them?

Apologies for the long post, feel free to skip over it if you have no interest in understanding how LFP cells die an early death

Bypassed the BMS allowing some of the cells to discharge to 0v, then put a battery charger on the whole battery at higher than an amp or so until the low cells reached above 3v .... and often reset the system by pressing the momentary relay latching switch as soon as the voltage is high enough to hold the contactor closed, and full solar and inverter charging also slammed into the cells ....... Then often leaving a dumb transformer type charger on the battery resulting in cell hitting 4.5v ..... all of these things will murder an LFP cell.

The over voltage charging on its own just increases the internal resistance, so when a cell reaches 3.6v, the charger is programmed to drop into float mode, but with the high internal resistance adds up to 0.5v to the actual cell internal voltage when charging and drops the voltage by the same amount when discharging ..... so the cell that cut off the charging @ 3.6v was actually only at 3.1v, then a load is applied and the voltage seen at the terminals is 2.6v ... and the BMS again isolates the battery ..... so they turn the bypass on again and yet another low voltage drain

Charging rapidly with the cell voltage below 1v will result in lithium actually transferring onto the graphite plate ... and that spot is always the spot where anything will attach to .... till it passes through the separator material and shorts out that plate in the cell ....
or
The rubbish that separates from the electrolyte coats the graphite material until no lithium ions can penetrate .... this results in the terminal voltage climbing rapidly till it trips the BMS, or gets that hot it boils off the electrolyte ..... No electrolyte, no method of transferring lithium ions from the active plate to the graphite plates ... so zero capacity, the same terminal voltage rapid climb and drop to 0v as a load is applied

T1 Terry

Everything degrades over time, it's just a matter of the rate.

Sure, but not the calendar drop shown in the graph previous and not the calendar shelf life of 10 yrs that is promoted across the interweb.

Manufacturers paid research labs big $$ to test their cells .... and they all use rapid aging techniques, then use the rapid cycling equivalent of 3 mths use (90 cycles) and extrapolate and expect cycle life from that .... LFP cells don't respond well to the rapid aging techniques, the electrolyte never has the chance to cool ... so it starts to breakdown ... resulting in rapid deterioration over short term, this false result multiplied by what ever they decided looked about right, and gave a cycle life based on that ......

The shelf life/use life is determined by the % capacity lost over time on a fully charged battery that is held at fully charged, then, with smoke and mirrors and using the page number and date into the equation, come up with 10 yrs .... yet ne explanation when queried on that number of yrs because they hadn't had any of the modern manufactured cell for 10 yrs to know ....

I guess you have all seen the reports that EV batteries are outlasting the lab test results that often gave a very short cycle life/kms travelled/age expectations, LFP have a better cycle life than NMC or any of the cobalt/lithium compounds, so treated correctly, they have a very long cycle/calander life.

T1 Terry

the theory is,that even black holes will eventually evaporate.

The other theories are the black holes eat each other, till there is one super massive black hole that can no longer find anything to feed itself ...... and it collapses ..... The Big Bang ....... and the cycle repeats ......

There is a theory, and well supported, that the solar system as we know it is the product of multiple repeats of a star collapsing in on itself .... until one such event resulted in the gas giants being in the wrong orbital position if compared to many other solar systems .... earth only has an abundance of carbon because of this, carbon along with a number of other elements, are formed in the centre of a star ......

T1 Terry

 

[decrep]

The other theories are the black holes eat each other, till there is one super massive black hole that can no longer find anything to feed itself ...... and it collapses ..... The Big Bang ....... and the cycle repeats ......

I like the bang crunch concept, but it seems to be out of favour at the moment.