MG4 51kWh LFP Battery Health

[Tight Git]

EU spec versions up to the "mild" facelift are rated for 117kW.

The early versions claimed 117kW charging ability but no-one has ever reported any rate achieved above 88kW so either a bit of marketing licence or MG realised they got it wrong and amended to the actual achievable rate.

 

[Peter WA]

These looks like the sort of cells that SAIC might be using. Dimensions look about right (they claim pack is 110mm thick) but their characteristics don't seem quite right.

Good news and bad news.

The bad: CATL doesn't sell cells on the open market AFAIK, and wouldn't sell to businesses that do. So these have either fallen off a truck, are recycled from wrecked cars, or plain old fakes. Which then begs the question how accurate the specs are.

The good: As to capacity tests: manufactures specify the discharge rate. For my home batteries EVE specified 0.5P. Yes, P, not C. Constant power, following the voltage curve, not constant current. And yes, those MB31 cells are rated at 314Ah and all tested in excess of 330Ah, so it is indeed possible to get more capacity than paid for.

 

[T1 Terry]

Interesting .....

It's certanly more plausible than them selling a battery that has actual capacity 10-20% lower than advertised. After all, MG4 is an export model and fooling overseas customers about something so essential like battery capacity just isn't on the table -- even for Chinese.

But fact of the matter is, battery manufacturing has variances, even high quality automotive cells. And typically, A tier manufacturer will rate their cells with the worst capacity that that series of cells is capable of.

This is true of any products that use lithium batteries. Look at phones, smartwatches, laptops etc.. Their cells usually have at two ratings; typical and minimum. And most often than not, a factory cell capacity will always be above typical.

iPhones for example use minimum capacity as their base 100% SoH. So an iphone with minimum rated capacity 2000mAh and 2050 typical capacity will likely have an actual capacity of ~2100mAh+ from the factory, but internal calculation will still show 100%. Only when capacity drops below 2000mAh, SoH starts dropping.

The early versions claimed 117kW charging ability but no-one has ever reported any rate achieved above 88kW so either a bit of marketing licence or MG realised they got it wrong and amended to the actual achievable rate.

So, fudging charge rates is ok, but creative maths on battery capacity just wouldn't happen ...... Check out the claimed battery capacity in later model MG4 51 models back early in the thread MG4 51kWh LFP Battery Health Might this be Saic covering their butt, a bit at a time
As Peter WA just posted as I one finger peck this post out through eyes that still aren't quite adjusted yet

As to capacity tests: manufactures specify the discharge rate.

So, what we need to know is, who manufactures the cells used in the MG4 51 and do they have capacity spec sheets available, so we can see how the 156Ah was determined ....

T1 Terry

 

[hojnikb]

The early versions claimed 117kW charging ability but no-one has ever reported any rate achieved above 88kW so either a bit of marketing licence or MG realised they got it wrong and amended to the actual achievable rate.

Could be. I'll test on mine, once it's done at the bodyshop. I'm really interested, because docs claim it can do it.

Interesting .....



So, fudging charge rates is ok, but creative maths on battery capacity just wouldn't happen ...... Check out the claimed battery capacity in later model MG4 51 models back early in the thread MG4 51kWh LFP Battery Health Might this be Saic covering their butt, a bit at a time
As Peter WA just posted as I one finger peck this post out through eyes that still aren't quite adjusted yet

So, what we need to know is, who manufactures the cells used in the MG4 51 and do they have capacity spec sheets available, so we can see how the 156Ah was determined ....

T1 Terry

No one said fudging any specs is ok, but i'm not sure 117kW was actually debunked as not being true. I'll have to test it out on mine, since it's an early built EU spec. So if any car can do it, this one could.


CATL is the manufacturer of MGs cells.

Unfortunally, i'm not convinced that they're lying about their capacity. The simple fact, that some people were able to hit over 400km on a single charge, proves that either it's the most efficient car in its class by far (even more than the old ioniq) which is unlikely given the aero or that the capacity is correct and the measurment, that people (like yourself) do, is not accurate.

 

[gdog]

OK, if 90% of the battery capacity was the amount added to the battery to bring it up from 10% to 100%, then that value represents 100% of the energy added. To get from 90% to 100%, 0.1 would equal 9% so multiplying the 90% by 1.1 adds 9% to the 90% = 99%. Add 0.01 of 90% = 0.9%. 90% x 1.11 = 99%

T1 Terry

Thanks for trying to enlighten me but I don't follow the logic TBH. My maths is pretty good but I don't understand the logic / process sorry.

Surely
100% = 90%+10%%+(some charging losses)%

 

[hojnikb]

Comparing an EV battery with a watch or phone battery, really? The Chinese are the masters of double speak, remain ambiguous by hiding behind language misinterpretations barriers ...... but that is a one way street. Having friend who is in direct business relations with the son of a major Chinese province, opens doors and also opens the eyes on the Chinese way of avoiding the truth, he is able to ask the number 1 son what is really meant from what was communicated, they are two very different things and it is the way some things are worded that makes all the difference.
The obvious double speak is put out as a decoy, you think you have spotted the ambiguity in what was said, the underlying word play is the part that comes back to bite you, simply because you didn't see the facts in what they said because it was hidden ... there in plain sight, if you knew what to look out for, but worded in a way that looked like what you wanted to see.

As an example, can you spot the double speak and implied message that is not actually stated anywhere? This is not a promotion or attack on this reseller, it is just an example
12V 100Ah LiFePO4 Deep Cycle Lithium ion Phosphate Battery 4WD RV Camping Marine | eBay

T1 Terry

Is phone battery fundamentally any different than a car battery? Yes chemistry is different is this particular case, but everything else applies just as well.

As for the battery you linked, i see nothing wrong. Everything they say is correct. Even capacity comparisment is on point, if compared to lead acid, as typical lead acid 100Ah battery will have half of it's capacity usable, unless you want to kill your battery very quickly.

The only "shady" thing about this particular battery is very poor BMS; it can only deliver 0,5C charge/discharge rate, while 1C+ being way more common on quality units.

And there's one thing we haven't discussed at all; What if T1s battery is defective?

I don't think he ever told us about the 0-100% range he's getting and what consumption he's getting. This might be a better starting point.

 

[siteguru]

He also intimated that he charged to 100% on a DC rapid charger ... I'm not sure that's even feasible. Maybe it was NMC chemistry but I'm sure I've seen more than one person here say that the charge capped at 97% when the charger stopped delivering charge. (Once you're above about 90% you get diminishing returns on a DC rapid charger - the charge rate plummets. You can generally do 10-80% again in the time it takes to go from 80-100% [if it ever gets to 100%]).

In terms of peak charge rate ... the early UK models were specified at 117kW but none ever achieved that. The maximum I recall anyone mentioning (with the LFP pack) is about 91kW. Note: whilst the peak specified charge rate was dropped from 117kW to 88kW, the actual specified 10-80% charge time remained unchanged. That's because this system holds the lower peak charge rate for a longer time vs the 117kW peak, so the overall power delivery time is the same.

 

[hojnikb]

Thanks for trying to enlighten me but I don't follow the logic TBH. My maths is pretty good but I don't understand the logic / process sorry.

Surely
100% = 90%+10%%+(some charging losses)%

Even if math is a little bit off, it still doesn't explain large discrepancy he's getting.

Even at 100% efficiency, the pack; if it was truly just 10% SoC at the time of charging, still isn't getting enough juice for the 51kWh nominal capacity.

So 3 options;

1) incorrect starting (or ending) SoC
2) inaccuracy on the DC charger when counting kWh (to the benefit of the customer)
3) defective battery having less capacity that it should have

These 3 options seem far more likely than MG lying about their capacity and getting amazing efficiency as a result.

He also intimated that he charged to 100% on a DC rapid charger ... I'm not sure that's even feasible. Maybe it was NMC chemistry but I'm sure I've seen more than one person here say that the charge capped at 97% when the charger stopped delivering charge. (Once you're above about 90% you get diminishing returns on a DC rapid charger - the charge rate plummets. You can generally do 10-80% again in the time it takes to go from 80-100% [if it ever gets to 100%]).

In terms of peak charge rate ... the early UK models were specified at 117kW but none ever achieved that. The maximum I recall anyone mentioning (with the LFP pack) is about 91kW. Note: whilst the peak specified charge rate was dropped from 117kW to 88kW, the actual specified 10-80% charge time remained unchanged. That's because this system holds the lower peak charge rate for a longer time vs the 117kW peak, so the overall power delivery time is the same.

That's in some ways better, especially on slower chargers. It's the curve shape or avarage 10-80% power that actually matters. But still, advertising some value, that isn't really achievable is really deceptive and should be called out. But then again, Tesla does the same thing with their "250kW" charge rate.

 

[king31]

The BMS SOC and the dashboard SOC are different. At 100% on the dashboard, the BMS SOC is at 99.9%. They remain identical until 88%, then a 1% discrepancy appears. For example:

• At 60% on the dashboard, the BMS SOC is around 61%.
• At 50% on the dashboard, the BMS SOC remains at 52%.
• At 17% on the dashboard, the BMS SOC is still 20%.
• At 10% on the dashboard, the BMS SOC is 13.5%.
• At 0% on the dashboard, the BMS SOC is still 4.5%, with the voltage finally dropping below 3.2V, around 3.12V to 3.14V depending on the battery temperature.

I gave you the answer, and you ignored it—I don’t understand you . The best thing to do is to get an OBD2 scanner and the CarScanner app—you’ll immediately get a clear answer to your questions. But to obtain accurate BMS information, the BMS software must be up to date.

Yes, there is a scheduled loss programming, and that annoys me too. Yes, our BMS does everything except properly balancing the cells at 100%, which gives me a headache when I see it working.


I even recorded its charging and balancing behavior, not via CarScanner (since that app wouldn’t be considered proof), but directly through the brand's software, VDS. You can admire its sloppy work... Even my cheap Chinese BMS does a better job at balancing.

I remind you that it’s a miracle that the minimum voltage finally exceeds 3.40V, so this result already satisfies me for now. I had to follow a slow and complicated procedure to allow the pack to balance so that the minimum cells could finally exceed 3.4V.

 

[hojnikb]

The BMS SOC and the dashboard SOC are different. At 100% on the dashboard, the BMS SOC is at 99.9%. They remain identical until 88%, then a 1% discrepancy appears. For example:

• At 60% on the dashboard, the BMS SOC is around 61%.
• At 50% on the dashboard, the BMS SOC remains at 52%.
• At 17% on the dashboard, the BMS SOC is still 20%.
• At 10% on the dashboard, the BMS SOC is 13.5%.
• At 0% on the dashboard, the BMS SOC is still 4.5%, with the voltage finally dropping below 3.2V, around 3.12V to 3.14V depending on the battery temperature.

I gave you the answer, and you ignored it—I don’t understand you . The best thing to do is to get an OBD2 scanner and the CarScanner app—you’ll immediately get a clear answer to your questions. But to obtain accurate BMS information, the BMS software must be up to date.

Yes, there is a scheduled loss programming, and that annoys me too. Yes, our BMS does everything except properly balancing the cells at 100%, which gives me a headache when I see it working.


I even recorded its charging and balancing behavior, not via CarScanner (since that app wouldn’t be considered proof), but directly through the brand's software, VDS. You can admire its sloppy work... Even my cheap Chinese BMS does a better job at balancing.
View attachment 35128

I remind you that it’s a miracle that the minimum voltage finally exceeds 3.40V, so this result already satisfies me for now. I had to follow a slow and complicated procedure to allow the pack to balance so that the minimum cells could finally exceed 3.4V.

View attachment 35129

How exactly is it not properly balancing?

 

[king31]

How exactly is it not properly balancing?

The BMS in our MG only removes excess energy to bring the voltages closer together until the delta is below 110 mV, then it cuts off the charge. It’s a really shaky charge management system. My Chinese BMS for my LFP solar battery does a much better job.

Then, all the cells drop to 3.33V (both min and max) as soon as the BMS SOC reaches 99%, staying that way down to around 30%, where the voltage changes to 3.28V. After that, it gradually decreases until 10%, where it reaches 3.20V. Finally, it keeps dropping until it reaches 3.1V / 3.13V at 0% on the dashboard, which corresponds to about 4.5% of the BMS SOC.

 

[hojnikb]

The BMS in our MG only removes excess energy to bring the voltages closer together until the delta is below 110 mV, then it cuts off the charge. It’s a really shaky charge management system. My Chinese BMS for my LFP solar battery does a much better job.

View attachment 35130
Then, all the cells drop to 3.33V (both min and max) as soon as the BMS SOC reaches 99%, staying that way down to around 30%, where the voltage changes to 3.28V. After that, it gradually decreases until 10%, where it reaches 3.20V. Finally, it keeps dropping until it reaches 3.1V / 3.13V at 0% on the dashboard, which corresponds to about 4.5% of the BMS SOC.
View attachment 35131

Oh, 110mV is really high! I do wonder why they selected such a high voltage.


As for how the balancing works, i'm not too surprised. AFAIK most EVs have passive balancers and as such waste excess energy when balancing.

I also have a home storage battery and i can easily balance my cells within 0,005V if needed. And it does so actively via the JKBMS.

My current voltages for example:

 

[king31]

Oh, 110mV is really high! I do wonder why they selected such a high voltage.


As for how the balancing works, i'm not too surprised. AFAIK most EVs have passive balancers and as such waste excess energy when balancing.

I also have a home storage battery and i can easily balance my cells within 0,005V if needed. And it does so actively via the JKBMS.

My current voltages for example:

View attachment 35132

If MG stopped charging at 3.45V, for example, the entire pack would be perfectly balanced, and the battery would last longer. Alternatively, they could push the charge limit up to 3.8V, like Tesla, to prioritize performance. At this voltage, all the cells would be slowed down, giving the weaker ones time to catch up.

Well, a new BMS update was released today, February 28, 2025, along with an IMCU update on February 19, 2025. I really hope they’ve improved charge management. I’m excited to test this new version! As soon as I do, I’ll come back here to share the observed changes.

 

[hojnikb]

If MG stopped charging at 3.45V, for example, the entire pack would be perfectly balanced, and the battery would last longer. Alternatively, they could push the charge limit up to 3.8V, like Tesla, to prioritize performance. At this voltage, all the cells would be slowed down, giving the weaker ones time to catch up.

Well, a new BMS update was released today, February 28, 2025, along with an IMCU update on February 19, 2025. I really hope they’ve improved charge management. I’m excited to test this new version! As soon as I do, I’ll come back here to share the observed changes.
View attachment 35133

Cool! I'm assuming you have access to the VDI interface and their online software?

 

[king31]

Super ! Je suppose que vous avez accès à l'interface VDI et à leur logiciel en ligne ?

Yes

 

[T1 Terry]

The BMS in our MG only removes excess energy to bring the voltages closer together until the delta is below 110 mV, then it cuts off the charge. It’s a really shaky charge management system. My Chinese BMS for my LFP solar battery does a much better job.

View attachment 35130
Then, all the cells drop to 3.33V (both min and max) as soon as the BMS SOC reaches 99%, staying that way down to around 30%, where the voltage changes to 3.28V. After that, it gradually decreases until 10%, where it reaches 3.20V. Finally, it keeps dropping until it reaches 3.1V / 3.13V at 0% on the dashboard, which corresponds to about 4.5% of the BMS SOC.
View attachment 35131

I'm assuming that is the computers assessment of the remaining capacity in the graph, or is it battery load in kW?
As I've said previously, you can't fully charge an LFP cell at 3.4V, the internal resistance will beat you every time. A rapid charge to 3.6V, then drop back to 3.45V, and allow an active balancer to do its thing will get an LFP battery back to 100% quickly and improve the cell longevity.

Those a cell temps I doubt we would ever see over here, to drag the cell temps that low would require a very low freezing point coolant, and that is never a good conductor of heat if it's water based, I'm assuming that the ambient temp was very low when those reading were taken ... ah, yes, I see the figure of 9°C in the left bottom corner

A lot of those readings still don't make a lot of sense, 3.7% of 51kWh is only roughly 9kWh remaining, the specs quoted previously say the min cell voltage a dragged down to 2.5V, a voltage between 3.1V and 3.13V does not represent a cell that has been dropped that low, nor does battery voltage of 323V, an average cell voltage of 3.1V, which looks more like the terminal voltage divided by the number of cells, rather than an actual cell voltage reading to come up with a min and max cell voltage ......

T1 Terry

Is phone battery fundamentally any different than a car battery? Yes chemistry is different is this particular case, but everything else applies just as well.

As for the battery you linked, i see nothing wrong. Everything they say is correct. Even capacity comparisment is on point, if compared to lead acid, as typical lead acid 100Ah battery will have half of it's capacity usable, unless you want to kill your battery very quickly.

The only "shady" thing about this particular battery is very poor BMS; it can only deliver 0,5C charge/discharge rate, while 1C+ being way more common on quality units.

The low continuous charge and discharge rates are pointer to the evaded truth about how the 100Ah capacity was measured. If they were honest and up front, they would have also listed the load testing regime used to support the 100Ah claim. A 50Ah @ C1 testing capacity would deliver 100Ah at a 1 amp or less load down to a rested voltage of 2.5V per cell average (specs say 10V).

The next implied but not actually stated thing in the description rambling, up to 10 x parallel capable, so that suggests 1,000Ah capacity, but also suggests the continuous discharge capacity would be 10 x 50 amps ..... but read through it again, that is never mentioned. The truth of the matter is, even 10 x 100Ah batteries in parallel, still only have a spec of 50 amps .... if any battery has to handle even a 100 amp load for more than 5 secs, the internal BMS fuse will fail and signals to the manufacturer, that the battery failure was because the specified parameters were exceeded.

Just another bit of deception by omission, no mention of the cell type used in the construction of the battery. The specs say 14.6V 100 amp peak for 5 secs and continuous charge rate of 50 amps, 3.65V per cell if the cells are perfectly balanced ....... how much heat would be generated attempting to burn off 182 watts (50 amps x 3.65V as suggested, but not stated, voltage per cell)? How would you dissipate that sort of heat, even if you could fit 4 x resistor that size, inside the battery .......

Looking at the selling price of the battery, what are the chances the BMS has the ability to lower the charge current ..... and how would it with no connection to the charger?

With a cut off voltage of 14.6V, the chance of any one of the 4 cells to go over 4V is very high .......

What are the chances of a 100 amp load close to the 10V cut off, not pulling a weak cell below 0v, even for 5 secs ......

There is lot of parameter monitoring required to go into a proper BMS ...... and they won't fit inside a battery case that is already full of LFP cells .....

T1 Terry

 

[Peter WA]

I also have a home storage battery and i can easily balance my cells within 0,005V if needed. And it does so actively via the JKBMS.

My current voltages for example:

I have the same set-up at home (also JK BMS), and yes, it is fantastic. And amazing how quickly it got the cells that well balanced. 2A active balancing is the way to go!

Careful though with the screenshot: at that voltage level cell voltages are very close together, even when cells aren't well balanced (like my ZS EV Mk2). I usually charge to 3.45V and balance there.

A couple of days ago I went to 3.55V, just to see what happens. With the active balancer operating, cell remained below 5mV mostly. FYI: there's about 0.4% capacity between 3.45 and 3.55 final charge voltage. Maybe abother 0.1% if you were to go to 3.65V. No point ever going there for our very well balanced home batteries. I'm happy to stay at 3.45V for my daily charging, 99.5% of capacity is plenty, and gives me a lot of headroom for over-voltages without getting into damage territory.

 

[hojnikb]

I have the same set-up at home (also JK BMS), and yes, it is fantastic. And amazing how quickly it got the cells that well balanced. 2A active balancing is the way to go!

Careful though with the screenshot: at that voltage level cell voltages are very close together, even when cells aren't well balanced (like my ZS EV Mk2). I usually charge to 3.45V and balance there.

A couple of days ago I went to 3.55V, just to see what happens. With the active balancer operating, cell remained below 5mV mostly. FYI: there's about 0.4% capacity between 3.45 and 3.55 final charge voltage. Maybe abother 0.1% if you were to go to 3.65V. No point ever going there for our very well balanced home batteries. I'm happy to stay at 3.45V for my daily charging, 99.5% of capacity is plenty, and gives me a lot of headroom for over-voltages without getting into damage territory.

Yeah, i keep my charging voltage at roughly 3.55V per cell and a bit more agressive balancing, because this pack is still very fresh and i'd like to really balance out and cycle the cells. Later i'll relax the settings and go for lower charging voltage of course.

It is difficult though, because this particular BMS i have can't be connected to the inverter (at least not with a middleware device converting signals). Seller claimed RS485 connection, but in reality it doesn't support it.

The low continuous charge and discharge rates are pointer to the evaded truth about how the 100Ah capacity was measure. If they were honest and up front, they would have also listed the load testing regime used to support the 100Ah claim. A 50Ah @ C1 testing capacity would deliver 100Ah at a 1 amp or less load down to a rested voltage of 2.5v per cell average (specs say 10v)

The next implied but not actually stated thing in the description rambling, up to 10 x parallel capable, so that suggests 1,000Ah capacity, but also suggests the continuous discharge capacity would be 10 x 50 amps ..... but read through it again, that is never mentioned. The truth of the matter is, even 10 x 100Ah batteries in parallel, still only have a spec of 50 amps .... if any battery has to handle even a 100 amp load for more than 5 secs, the internal BMS fuse will fail and signals to the manufacturer, that the battery failure was because the specified parameters were exceeded.

Just another bit of deception by omission, no mention of the cell type used in the construction of the battery. The specs say 14.6v 100 amp peak for 5 secs and continuous charge rate of 50 amps, 3.65 per cell if the cells are perfectly balanced ....... how much heat would be generated attempting to burn off 182 watts (50 amps x 3.65v as suggested, but not stated, voltage per cell) How would you dissipate that sort of heat, even if you could fit 4 x resistor that size, inside the battery .......

Looking at the selling price of the battery, what are the chances the BMS has the ability to lower the charge current ..... and how would it with no connection to the charger?

With a cut off voltage of 14.6v, the chance of any one of the 4 cells to go over 4v is very high .......

What are the chances of a 100 amp load close to the 10v cut off, not pulling a weak cell below 0v, even for 5 secs ......

There is lot of parameters monitoring required to go into a proper BMS ...... and they won't fit inside a battery case that is already full of LFP cells .....

T1 Terry

A cheap sealed battery like this will not give you exact data on what BMS it uses, what testing was done to give that capacity rating or anything else really. That's just isn't the norm, even with higher quality batteries.

The one thing i'd worry about a battery like this is really just the low temp disconnect (a lot of cheap ones don't have it). I'm sure there's a balancer inside and low/high cell voltage disconnect, so no cell will ever go down to 0V or go up to 4V, unless something seriously goes wrong with the BMS. Even the shittiest of shitty BMSs have low/high voltage protections.

And i don't really get your point about balancing. Why would you need to dissipate 182W of heat?
Passive balacing doesn't work like that...

 

[T1 Terry]

At 3.34V, I would have expected them to all be within a few millivolts. LFP cells, if they are within 50 millivolts at 3.45V, don't go out of balance until they drop below 3.0V, unless there is a serious high resistance connection between series connected cells ..... or the voltage sensing connection of course.

T1 Terry

A cheap sealed battery like this will not give you exact data on what BMS it uses, what testing was done to give that capacity rating or anything else really. That's just isn't the norm, even with higher quality batteries.

The one thing i'd worry about a battery like this is really just the low temp disconnect (a lot of cheap ones don't have it). I'm sure there's a balancer inside and low/high cell voltage disconnect, so no cell will ever go down to 0V or go up to 4V, unless something seriously goes wrong with the BMS. Even the shittiest of shitty BMSs have low/high voltage protections.

And i don't really get your point about balancing. Why would you need to dissipate 182W of heat?
Passive balacing doesn't work like that...

How do you think passive balancing works? If you used a charger that was within the specs they quote for the battery, 50 amps and 14.6V, how do you burn off the high cell voltage to allow the low cell voltages to catch up?
These cheap internal BMS are just basic bicycle battery balancers, designed for 5 Ah or less cells, they rely on terminal voltage and all cells being within near perfect balance .... at an average end of charge voltage of 14.6V, it requires no differential between highest and lowest cell to maintain the 3.65V max, if there is any deviation, the 50 amps x 3.65V = 182.5W, is required to burn off the incoming charge current to that cell will the other 3 cells in the battery catch up .......

T1 Terry

 

[Peter WA]

It is difficult though, because this particular BMS i have can't be connected to the inverter (at least not with a middleware device converting signals). Seller claimed RS485 connection, but in reality it doesn't support it.

You've got the JK Inverter BMS? That's the one I have. It happily talks with my Victron Multiplus II 48/5000 Inverter Charger via CAN bus. I run it in close-loop configuration (BMS is in the driver's seat, Inverter does what it is told with respect to target voltage and current) and have not had any issues yet.