• We are having a problem with new Hotmail members being unable to receive activation emails. Please avoid using a Hotmail email address. Thanks.

MG ZS EV MK2 - LFP (standard range) - Battery SOH Theory

Peter WA

Established Member
Joined
Jan 18, 2021
Messages
102
Reaction score
110
Points
44
Location
Perth, Australia
Driving
MG ZS EV
I've got a theory that I would like to test.

The theory is: in the Mk2 ZS EV with LFP battery (standard range), the battery State of Health (SOH) reported by the car is a fudge, that is nothing other than a count-down timer, that started running when the battery pack was manufactured, and decreases linearly with time.

That countdown is calibrated to ensure that it will remain over 70% for the duration of the 7-year warranty period. The decrease in reported SOH is set to about 3.8% per year, which over 7 years would equate to 26.6%. Or 0.0104% per day.

Here is a chart of my car's SOH since I got OVMS working for it, about 1 week after I got the car. The only bumps are when the data wasn't updated for a few days. Feel free to hold a ruler against the bottom of the line to see how straight it is.

SOH linear Nov22 to Jan23.png



So far I've only got data for 2 cars. My own and one other. For my car: production date of the car is known, according to contract information: 28/8/2022. If I extend the chart back into the past, it will reach 100% on 18/8/2022 - quite reasonable to have the battery pack assembled 10 days before the vehicle left the factory.

It would be great if some people here can confirm (or refute) that theory. Please post SOH and production date as replies. Only Mk2 ZS EV with LFP battery please (standard range). Or alternatively, if you don't know production date, read out SOH a few days apart, and check whether the daily decline is in line with what I predict.

Of course I've got an explanation. I believe MG engineers found it too hard to measure SOH. And a meeting took place a bit like this (all facts above, pure fiction below):

Product manager: Did you finally manage to measure SOH for the new LFP batteries?
Engineer: Sorry, we can't figure out how. The voltage curve is just too flat. Maybe we'll have some great idea later, or it will become easier to measure once there's significant degradation.
Product manager: But the cars will enter production next month, and SOH is part of the warranty conditions. We need to populate that PID.
Engineer: How about we just set it to 100% and once we've figured out how to measure, we'll apply a software update?
Product manager: No one will believe it stays at 100%, if we do that, at the first service people will find out something isn't quite right.
Engineer: How about a linear decrease over time then?
Product manager: Sounds like a plan, just make sure it stays above the warranty cut-off. And make sure you keep working on it, so we can do that software update to fix it.
 
If my theory is correct, the SOH reported by the car is useless. Which is a good thing, I would have been very disappointed if my car's battery SOH was only a tad over 99% the day I received it, with 7km on the odometer. Other people got cars with longer journeys (held up in customs), and had only 98% SOH when they received them.

I believe the actual SOH is a lot higher still, and for most owners will decrease a lot less than the linear decrease that is built into this fudge. I believe the hardware is great, but the software dodgy.

2 problems in different scenarios:

If a buyer checks battery SOH before buying a 2nd hand vehicle, under-reporting SOH will needlessly reduce the price that can be achieved when selling the car.

If a battery actually is faulty, and dealers (who in general are totally clueless) just read SOH, they will say: "No warranty case, battery is fine, we've checked the SOH. The low range you're reporting comes from how you drive."
 
I've been tracking I can share my data when I'm home, but what about other users that are still 100%
 

Attachments

  • IMG_20230118_003032_509.jpg
    IMG_20230118_003032_509.jpg
    29.7 KB · Views: 157
I've got a theory that I would like to test.

The theory is: in the Mk2 ZS EV with LFP battery (standard range), the battery State of Health (SOH) reported by the car is a fudge, that is nothing other than a count-down timer, that started running when the battery pack was manufactured, and decreases linearly with time.

That countdown is calibrated to ensure that it will remain over 70% for the duration of the 7-year warranty period. The decrease in reported SOH is set to about 3.8% per year, which over 7 years would equate to 26.6%. Or 0.0104% per day.

Here is a chart of my car's SOH since I got OVMS working for it, about 1 week after I got the car. The only bumps are when the data wasn't updated for a few days. Feel free to hold a ruler against the bottom of the line to see how straight it is.

View attachment 14552


So far I've only got data for 2 cars. My own and one other. For my car: production date of the car is known, according to contract information: 28/8/2022. If I extend the chart back into the past, it will reach 100% on 18/8/2022 - quite reasonable to have the battery pack assembled 10 days before the vehicle left the factory.

It would be great if some people here can confirm (or refute) that theory. Please post SOH and production date as replies. Only Mk2 ZS EV with LFP battery please (standard range). Or alternatively, if you don't know production date, read out SOH a few days apart, and check whether the daily decline is in line with what I predict.

Of course I've got an explanation. I believe MG engineers found it too hard to measure SOH. And a meeting took place a bit like this (all facts above, pure fiction below):

Product manager: Did you finally manage to measure SOH for the new LFP batteries?
Engineer: Sorry, we can't figure out how. The voltage curve is just too flat. Maybe we'll have some great idea later, or it will become easier to measure once there's significant degradation.
Product manager: But the cars will enter production next month, and SOH is part of the warranty conditions. We need to populate that PID.
Engineer: How about we just set it to 100% and once we've figured out how to measure, we'll apply a software update?
Product manager: No one will believe it stays at 100%, if we do that, at the first service people will find out something isn't quite right.
Engineer: How about a linear decrease over time then?
Product manager: Sounds like a plan, just make sure it stays above the warranty cut-off. And make sure you keep working on it, so we can do that software update to fix it.
Product engineer: and let’s set the LKA at dangerous always-on levels - that should distract them from the battery health! Oh sorry, that’s for the MG4 Meeting.
 
Product engineer: and let’s set the LKA at dangerous always-on levels - that should distract them from the battery health! Oh sorry, that’s for the MG4 Meeting.
I think rather than distraction they might have been aiming for outright death.
 
I've got a theory that I would like to test.

The theory is: in the Mk2 ZS EV with LFP battery (standard range), the battery State of Health (SOH) reported by the car is a fudge, that is nothing other than a count-down timer, that started running when the battery pack was manufactured, and decreases linearly with time.

That countdown is calibrated to ensure that it will remain over 70% for the duration of the 7-year warranty period. The decrease in reported SOH is set to about 3.8% per year, which over 7 years would equate to 26.6%. Or 0.0104% per day.

Here is a chart of my car's SOH since I got OVMS working for it, about 1 week after I got the car. The only bumps are when the data wasn't updated for a few days. Feel free to hold a ruler against the bottom of the line to see how straight it is.

View attachment 14552


So far I've only got data for 2 cars. My own and one other. For my car: production date of the car is known, according to contract information: 28/8/2022. If I extend the chart back into the past, it will reach 100% on 18/8/2022 - quite reasonable to have the battery pack assembled 10 days before the vehicle left the factory.

It would be great if some people here can confirm (or refute) that theory. Please post SOH and production date as replies. Only Mk2 ZS EV with LFP battery please (standard range). Or alternatively, if you don't know production date, read out SOH a few days apart, and check whether the daily decline is in line with what I predict.

Of course I've got an explanation. I believe MG engineers found it too hard to measure SOH. And a meeting took place a bit like this (all facts above, pure fiction below):

Product manager: Did you finally manage to measure SOH for the new LFP batteries?
Engineer: Sorry, we can't figure out how. The voltage curve is just too flat. Maybe we'll have some great idea later, or it will become easier to measure once there's significant degradation.
Product manager: But the cars will enter production next month, and SOH is part of the warranty conditions. We need to populate that PID.
Engineer: How about we just set it to 100% and once we've figured out how to measure, we'll apply a software update?
Product manager: No one will believe it stays at 100%, if we do that, at the first service people will find out something isn't quite right.
Engineer: How about a linear decrease over time then?
Product manager: Sounds like a plan, just make sure it stays above the warranty cut-off. And make sure you keep working on it, so we can do that software update to fix it.
It's certainly plausible that they might do this but I also suspect that real degradation would also follow this pattern when viewed over such a small time period.

Have you witnessed a change in SOH without charging or does the change always follow a charge?
 
I've been tracking I can share my data when I'm home, but what about other users that are still 100%

Any standard range MK2 (with LFP battery) still at 100%?

That would get in the way of my theory. I'd ask about software versions next, perhaps there is an update already that fixes the anomaly, but my dealer (and dealers for others who report low SOH) failed to install it.

I think rather than distraction they might have been aiming for outright death.
Maybe just for extra sales of replacement cars. Enough airbags that the occupants should be fine, but the vehicle written off ;-)

It's certainly plausible that they might do this but I also suspect that real degradation would also follow this pattern when viewed over such a small time period.

Have you witnessed a change in SOH without charging or does the change always follow a charge?

The decline is independent of driving distance / charging history.

For the value to update, the car has to be 'on' for a moment. But even a 2km trip to the post box will then result in the same daily decline as a 200km trip with charging.

About half the total distance driven (and charging) was in that single 10 day bit in the middle of the chart where no updates were recorded in my database (because the car wasn't home on WiFi). I did a 2000km roadtrip. Once back, the decline was perfectly in line with before, just making up for missing days, not affected by all the km or charging.
 
The decline is independent of driving distance / charging history.

For the value to update, the car has to be 'on' for a moment. But even a 2km trip to the post box will then result in the same daily decline as a 200km trip with charging.

About half the total distance driven (and charging) was in that single 10 day bit in the middle of the chart where no updates were recorded in my database (because the car wasn't home on WiFi). I did a 2000km roadtrip. Once back, the decline was perfectly in line with before, just making up for missing days, not affected by all the km or charging.
I suppose you counter this by saying that LFP is not expected to degrade much due to cycles, most of it will be due to calendar aging instead.

You could look for differences between cars in different climates but that still wouldn't rule out MG using a temperature factor in the calculation.

A degradation rate of 3.8% in the first year seems quite good so there is nothing to worry about yet. This is a graph showing 9 months of storage in one study:

1674030642885.png


Your graph is suspiciously linear though. Perhaps they use an estimation as you suggest but hopefully it updates with real values every so often. The same graph from my LEAF over 8.5 years shows much more variation as you would expect if it was using a real measurement:

1674030958176.png


This thread could get really interesting once more people post data covering a longer time period.
 
I've been tracking I can share my data when I'm home, but what about other users that are still 100%


Only just saw your attachment. The measured values line up perfectly with my prediction: 22/12/23 to 15/1/23 are 24 days. Calculated decline in SOH: 24 * 0.0104% = 0.25%. Actual decline in SOH: 96.82% - 96.57% = 0.25%.

I assume you set the day you go the car to 100%, but didn't actually check SOH back then?

My calculation would go: 96.57% is a loss of 3.43%, equal to 330 days of linear decline. I'd put the production day of you battery pack at 19/02/2022.

Do you have a sticker on the driver's side door of your car, with year and month on it? That's when it rolled out of the factory. Battery pack would have been made some days or weeks or even months before that.

@MoDolph Edited - took a little to understand that chart. So that's looking at decline in capacity for a battery stored at a given SOC. Great data, certainly supports my argument that even LFP should not be routinely charged to 100%. Surprised to see how well it supposedly worked being stored at very low SOCs, that's contrary to data I had seen before that saw quite some damage caused by SOC below 20%.

I agree, this will be interesting, hoping that a few people have data. @lusitano posted some data, and it seems to align very well with my predictions, despite taking place at a very different SOH, their car is already below 96%, while my analysis was around 99%.
 
Last edited:
@MoDolph Your first chart is state of charge when a full battery is stored. Measuring self-discharge over time. Very different concept to actual decay in capacity when fully charged.
That is incorrect. It is the capacity fade after 9 months for various storage SOCs. I've attached the full paper.
 

Attachments

  • Keil_2016_J._Electrochem._Soc._163_A1872.pdf
    798.4 KB · Views: 142
Yes, as didn't have way to measure, when i got it, but as I uunderstood my car sat for months somewhere before I bought it as the MG warranty sticker lost its colour from sun. Will take some more measurements tomorrow, not had time today, but I think you right the drop happens even if car doesn't move.
 
Support us by becoming a Premium Member

Latest MG EVs video

New EVs from MG: MG S9 & MG9 plus hot topics from the forums
Subscribe to our YouTube channel
Back
Top Bottom