Reading SOH

[MarkHoward]

Has anyone discovered a reliable way of reading the SOH in the MG4?
I have a VPeake ODB dongle.

Can't believe that MG has not provided this figure natively through the Infotainment system.
My Previous 7 years older BMW i3, did.

 

[EVentropy]

Well, it went into trickle charge mode, before I could see the current at transition. Now a relay is chicking on an off, I guess as it tops up to balance low cells. Obviously not re- distribute as in capacitive balancing that cuts ext charger off completely and takes from high cells.
Well, that lasted all of five minutes!
Comment welcome on my understanding of this, or lack of.

 

[EVentropy]

Ok, last post.
Supposedly I have finished balancing. CS table all sensors data max cell voltage = 3.49V; min = 3.34V. Delta = 0.15V
My house bank after balancing is ~ 3mV
My house bank before balancing = -0.15V
ADJUSTABLE. For start of balancing,
Hmmmmm!

 

[Coulomb]

Point being below, info centre says power to battery has reduced from 1.4 kW to one kW yet clamp is still reading 6.6A.

That's because other things (computers, pumps, charging the auxiliary (12 V) battery) are changing loads, and MG chooses to show the net power into the HV battery. In particular, auxiliary battery charging will be heavy early on, and will taper as time progresses. When charging at a low rate like the granny "charger", this will cause noticeable changes, as you noted.

Edit: But I just noticed that you are at 100% and therefore balancing. It's not obvious to me whether they consider balancing current (which is probably burned off as heat) is part of the "net current into the battery". I would have thought so, but it looks like it may not.

 

[Everest]

Under load, a weak cell will show up instantly, a high resistance cell will show up when charging, a strange delta v figure can be traced to the problem area. All these things affect the battery operation, fast charging stops early if a cell shows signs of high resistance, the same for the remaining battery capacity, better to know and put the problem in writing to the dealership, than be told the battery is at 94% SOH when you know it isn't .....

absolutely. But my point was you would have that type of data with just the min/max cell # and voltage - the average user doesn't need all cells to spot a problem. You or I, though are not average users and we like to have all the data

Point being below, info centre says power to battery has reduced from 1.4 kW to one kW yet clamp is still reading 6.6A. ( on granny charger of course)

As @Coulomb mentioned, there are other things taking the load.

My calculation of 6.6A @ around 230V is only 1500W. Take out 200W quiescent usage of car's electronics during charging (I got that figure from reading about the Ioniq 5, but guess the MG4 is similar amount), then take out 10% for inverter efficiency (another 130W) and some power losses along the cables, then 6.6A doesn't sound wrong to me for 1000W into the battery itself.

Supposedly I have finished balancing. CS table all sensors data max cell voltage = 3.49V; min = 3.34V. Delta = 0.15V

Interesting. So, from that info, I'd conclude that the MG4's LFP BMS is taking the max cell to 3.5V as a measure of 100% SOC. That makes perfect sense to me, any higher and longevity would be impacted for a very very small percentage increase in stored power.

However, lowest cell at 3.34V is not good IMHO. I've no experience of car batteries, but (as you mentioned) on a home / boat energy storage battery, I'd be expecting only a few 10's of mV delta max - guess it is easier to achieve with (say) 16 cells, than 104 of them though!

 

[EVentropy]

Coulomb- Possible, but hard to be convinced when power into HV bank drops 0.4kW and current doesn’t move from previous. It had been charging for some 20hrs. 6.6A is all the granny charger can supply so what was supplying pump ,electronic etc - the aux battery as it was still blowing off its surface charge at 13.8V. Speculation really as individual devices are not monitored, or we are not privileged the data!

“Everest—Interesting. So, from that info, I'd conclude that the MG4's LFP BMS is taking the max cell to 3.5V as a measure of 100% SOC. That makes perfect sense to me, any higher and longevity would be impacted for a very very small percentage increase in stored power.

However, lowest cell at 3.34V is not good IMHO. I've no experience of car batteries, but (as you mentioned) on a home / boat energy storage battery, I'd be expecting only a few 10's of mV delta max - guess it is easier to achieve with (say) 16 cells, than 104 of them though!”

Absolutely All you need is one dragging it feet.
When it went into balance it was pulsing 2.4A down to 0.4A then back to 2.4A on clamp.
Relay controlled balancer? Doesn’t sound right to me. I am NOT comfortable with this experience and are talking to technicians tomorrow. Pity it took me so long to get my A into G and watch a charge completion with unverified data to wave a flag. I assumed all was well. Cheers.

 

[Everest]

Possible, but hard to be convinced when power into HV bank drops 0.4kW and current doesn’t move from previous. It had been charging for some 20hrs. 6.6A

Ah.. I didn't realise it hadn't changed. But, as you say, we don't know what really goes on under the covers.. one reason I implemented my own control systems for my house battery! Not really practical to do that for an EV though.

When it went into balance it was pulsing 2.4A down to 0.4A then back to 2.4A on clamp.
Relay controlled balancer? Doesn’t sound right to me.

I have no idea, but with the large current draws of an EV, I would expect the car's BMS to use a contactor rather than the MOSFET's that my home BMS uses. A FET-based BMS would need a lot of heatsink and wasted energy and more likely to fail IME.

So, I can imagine a situation where the BMS cuts out when one cell reaches 3.5V, then shortly after, the voltage on that cell reduces (to whatever limit is set), then the contactor closes and more charge is added, until 3.5V is reached again. Hopefully, every time that happens the balancer is doing its job, whether its a passive or active balancer, so over time, the cells will get closer in SOC. But who knows? I am only guessing

 

[Coulomb]

When it went into balance it was pulsing 2.4A down to 0.4A then back to 2.4A on clamp.
Relay controlled balancer?

If by "relay" you mean a surface mount or even on-chip Field Effect Transistor, then yes, that's possibly what's going on. Typically when the BMS notices a cell is too high in voltage and balancing is in effect, it sends commands to the voltage monitoring chip (which typically covers 8-12 cells at once) to turn on the internal transistor that connects an external balance resistor across the cell. This external resistor can be a fairly high value, e.g. early Nissan Leafs used 430Ω for a bypass current of less than 10 mA. Often an external FET is used to allow higher balance currents. The bypass current could be of the order of 100 mA, which with a 400 V battery comes to some 40 W if every balance resistor was on (which makes no sense, but all except one could make sense if one cell was low in voltage). I'll admit that 400 W of bypass current is hard to imagine; there would be way too much heat way too close to the cells.

I'm not aware of anyone analysing any MG Battery Management Systems in the same level of detail that iMiev, Leaf, and Tesla BMS have been. But I've not been looking either.

Edit: The 400 W changes in AC charge current is presumably due to the On-Board charger sending alternately 0 W and 400 W to the main battery. Bypass current is typically 1-10% of the charge current, but the exact values can vary widely, I believe.

Imagine one cell getting too high in voltage. The charge current is set to 0. After a while, the voltage delta falls below a threshold, so the BMS instructs the On Board Charger to apply a very modest 400 W (some 100 mA into say 156 Ah cells, roughly 0.0064C or C/1560 (C being the capacity of the cell in Ah, used as a current. 1C current will completely charge or discharge a battery in one hour.) After another while, the delta passes another threshold, and the BMS has to tell the OBC to stop charging again. This repeats while the low voltage cell gradually comes into balance. Depending on how unbalanced the cells are, this could happen faster or slower, and at higher or lower charge power. When the low voltage cell becomes much closer to the average cell voltage, the OBC might be instructed to increase the charge power to thousands of watts, until all the cells are full.

Often this would happen in reverse if a cell was too high in voltage. Charging at say 98% could be at full power, 99% SoC could be at say 4 kW, then pulled way back to say 400W as the first cell starts getting too full.

 

[Everest]

C being the capacity of the cell in Ah, used as a current. 1C current will completely charge or discharge a battery in one hour.

that so confused me a few years ago until I realised 'C' in that context did not mean the Coulomb!

 

[EVentropy]

"imagine one cell getting too high in voltage. The charge current is set to 0. After a while, the voltage delta falls below a threshold, so the BMS instructs the On Board Charger to apply a very modest 400 W (some 100 mA into say 156 Ah cells, roughly 0.0064C or C/1560 (C being the capacity of the cell in Ah, used as a current. 1C current will completely charge or discharge a battery in one hour.) After another while, the delta passes another threshold, and the BMS has to tell the OBC to stop charging again. This repeats while the low voltage cell gradually comes into balance. Depending on how unbalanced the cells are, this could happen faster or slower, and at higher or lower charge power. When the low voltage cell becomes much closer to the average cell voltage, the OBC might be instructed to increase the charge power to thousands of watts, until all the cells are full."

Having first introduction to LFP and BMS's with fully user configurable thresholds, I was hoping for little imagining. ie Voltage to start balance, voltage delta to start balance, HV disconnect charge bus, Voltage to resume charge, well documented with understandable nomenclature and FET's. (that are silent.) I hear you Everest. Heat and FETs.
I have just a vague idea with the MG4 LFP as its all in stealth mode, except for the noise from contactors!
Its like turning on the light with a fully featured UI, and adjustable BMS parameters.
Yes, a somewhat dubious business model in terms of safety. But the curious (or pissed off) will hack it anyway in the end. Nobody wants to be told what they can't have or what they shouldn't have, anymore.

Diagnostic VDI, VNCI, MDI with MG4

[ Edit moderator: Links to various useful posts for quick access. Mention @Coulomb to have more links added. A tutorial guide for software updates. Trophy/Luxury 22/23 FICM loading guide (separate thread). The various SAIC server access plans and costs. An inexpensive VDI dongle needed for...

www.mgevs.com

 

[T1 Terry]

I have a laugh to myself every time I see "bypass resistor," face it, that resistor is a controlled short across the cell, nothing more. That energy is converted to heat, a loss system simply creating heat in an area that is heat sensitive .... cheap but nasty

Why haven't dynamic balancers been adopted by battery manufacturers? An induction coil, charged by the cell, then the cell link is cut and all the coils linked, then the coil link broken and cell link reactivated.
The principle used is the higher voltage cell dumps more current into the coil than a low voltage cell, when they link together, all the coils balance out the voltage and therefore the capacity, then they link back to the respective cell. If the voltage in the cell is lower than the voltage in the coil, current flows into that cell, if the cell voltage is high, some of the current flows into the coil till the voltage equalises between the cell and the coil, the system repeats until all the cell voltages are within what ever spec is programmed in, then the whole thing shuts down .... until a voltage delta exceeds the switch on point, and it starts again.

T1 Terry

 

[Coulomb]

Why haven't dynamic balancers been adopted by battery manufacturers?

I haven't studied this closely, but I suspect that the problem is scaling this idea up to a hundred or so cells. Twice that for an "800V" battery.

 

[Everest]

Why haven't dynamic balancers been adopted by battery manufacturers?

Surely EV BMS's use active balancers (capacitive, rather than induction coil), rather than passive?

I wouldn't have thought passive balancers would be suitable, given the short time cells are charged up to 100%?

 

[Coulomb]

I wouldn't have thought passive balancers would be suitable, given the short time cells are charged up to 100%?

That's a good point. The Nissan Leaf with its very low <10 mA balancing supposedly is balancing all the time (possible with NMC I guess). And 10 mA won't drain a 60+ Ah cell pair for a looong time.

It would be good to see inside one.

I thought with capacitive, you can only pass the charge on to your neighbouring cells. So it might take a long time and/or be inefficient if you have to shuffle charge a long way (e.g. low and high cells at nearly opposite ends of the pack).

 

[T1 Terry]

Surely EV BMS's use active balancers (capacitive, rather than induction coil), rather than passive?

I wouldn't have thought passive balancers would be suitable, given the short time cells are charged up to 100%?

I'm sure to be corrected here, but from memory, capacitors can only move as much as they can hold in the time they have to charge and discharge. An induction coil can be saturation charged much faster, doesn't deteriorate as fast as a capacitor and can dissipate the stored current very quickly, therefore can be cycled faster, thus able to move a greater amount of cell capacity over a given time period.
The catch is, when using LFP or LYP cells, that difference between 100% SOC and less than 100% SOC is only seen at the top end of the charging cycle. This means the charging needs to recommence frequently to bring the cell at 100% capacity up to the charge cut voltage to create enough voltage differential for capacity to move from the full cell to the not so full cells.

SOC also needs to be understood, this is when the cell can not store any more capacity and this causes the voltage to rise .... this can be caused by a number of factors, a higher internal resistance so the high cell voltage is seen before the cell actually reaches saturation charged, a cell with less capacity than the others, a poor connection between cells resulting in some of the charge current being wasted as heat through the high resistance joint .... and with LFP cells, the day of the week or page number or something .... a random unknown seems to sneak in there some where that will be an issue today but not seen again for weeks or even mths ......

The other thing to keep in mind, if a cell reaches saturation charged but that capacity doesn't get moved to the other cells fast enough to beat the charger dropping to a lower charge voltage, each cycle will get that cell ever higher at the expense of the other cells dropping ever lower, so a decent amount of capacity needs to move each cycle if the cell capacity is significant ...... it takes a lot longer to balance a 125Ah cell than an 18650 1000mah cell, the type used for making those fantastic graphs used to show how great their cell balancer is ......

T1 Terry

 

[Everest]

I thought with capacitive, you can only pass the charge on to your neighbouring cells.

Not IME. They use solid state switches to pass charge from highest to lowest voltage cell. They extract electrons from the highest voltage cell and store the charge in 2 capacitors. Then the 2 capacitors are connected in series and discharged into the lowest cell.

Coulombs and all that, @Coulomb

 

[EVentropy]

Ok, last post.
Supposedly I have finished balancing. CS table all sensors data max cell voltage = 3.49V; min = 3.34V. Delta = 0.15V
My house bank after balancing is ~ 3mV
My house bank before balancing = -0.15V
ADJUSTABLE. For start of balancing,
Hmmmmm!View attachment 33019

Quick update clarification on this. Battery reached 99% SOC, then it went into "trickle charge" for a couple of minutes to 100%. Then contactor started cycling for~5min. I'm not sure when above table was a reflection of.
I used the car, to 83% then did another data grab and it showed delta of < 10mV Min to max in line with my other battery experience. (CALB 230Ahr boat batteries. LiFePO4 4S2P 2x bms, very configurable)
Next 100% charge I will do multiple grabs as it is balancing. Unless C. scanner does real time graphs until unplugged.
which I doubt for a free app.
Edit: What I really want to understand is the mode of balancing as my house install is capacitive and doesn't require external power. Just linking high cell to low cell in parallel.,
sequentially. until it averages out. Is this passive?
BTW, the clacking contactors is the MG way of deciding that the battery is 100%, according to MG tech's. somewhat primitive I would have thought?

Too much to learn, too little time. The reason I need to know, is there is no "help" readily available 100's of miles offshore.

 

[kenneth104]

Capacitor-inductor or DCDC active balancing, their cost and complexity have greatly increased. Some early Chinese vehicles (using OptimumNano batteries) also had active balancing BMS, but they could not reverse the problem of large differences in monomers.

So most of the current vehicles are simple bypass resistors

 

[Everest]

I used the car, to 83% then did another data grab and it showed delta of < 10mV Min to max in line with my other battery experience. (CALB 230Ahr boat batteries. LiFePO4 4S2P 2x bms, very configurable)

That would be expected. Unless cells are above about 3.4V they will be at a similar voltage to each other, even with different SOC. See also my graph here in the LFP battery health thread...

MG4 51kWh LFP Battery Health

If anyone sees any other errors, please point them out My 2p worth... The 51kWh is actually a rounded up figure. The 400v can only be achieved by charging the 104 cells to 3.85V ..... Any cell voltage above 3.4V is surface charge, put it under load and that 0.45VDC over charge will vanish...

www.mgevs.com

What I really want to understand is the mode of balancing as my house install is capacitive and doesn't require external power. Just linking high cell to low cell in parallel.,sequentially. until it averages out. Is this passive?

No, that is active. Passive is the bleeding off of excess energy through a resistor - i.e. using the energy to heat the resistor up.

 

[T1 Terry]

Quick update clarification on this. Battery reached 99% SOC, then it went into "trickle charge" for a couple of minutes to 100%. Then contactor started cycling for~5min. I'm not sure when above table was a reflection of.
I used the car, to 83% then did another data grab and it showed delta of < 10mV Min to max in line with my other battery experience. (CALB 230Ahr boat batteries. LiFePO4 4S2P 2x bms, very configurable)
Next 100% charge I will do multiple grabs as it is balancing. Unless C. scanner does real time graphs until unplugged.
which I doubt for a free app.
Edit: What I really want to understand is the mode of balancing as my house install is capacitive and doesn't require external power. Just linking high cell to low cell in parallel.,
sequentially. until it averages out. Is this passive?
BTW, the clacking contactors is the MG way of deciding that the battery is 100%, according to MG tech's. somewhat primitive I would have thought?

Too much to learn, too little time. The reason I need to know, is there is no "help" readily available 100's of miles offshore.

The SOC % is meaningless until all the cells are 3.5V or higher. All the 100% is saying is it sees the voltage it expects to see when 100% SOC, or it is a tad out of wack as far as calibration and has not accurately counted the energy out V energy in, and thinks more has come in than actually has, or hasn't recorded some of what went out.
The last one is common if the cell voltage monitor doesn't take the battery negative from the load side of the energy counting shunt, it doesn't see the energy the monitor takes to power itself being drawn out of the battery, so it doesn't get recorded.

The battery is only at 100% SOC when all the cells are above 3.5V, if any cell isn't above 3.5V then the battery isn't fully charged.

Two methods of cell balancing:
If only 1 cell group is low, a single 3.6V charger powered by the inverter will use energy from the whole pack and feed it into the low cell the charger is connected across. Leave the full battery charging system running while this balancing is taking place ..... if the other cells drop below 3.5V, you now have no idea what is fully charged, you could end up with the cell you are trying to bring up to the same voltage as the other becoming the high voltage cell once a full battery charge takes place, the cell you are trying to bring up might only need a few Ah to get the voltage up, put too many Ah in and that cell will reach 100% SOC before the others, creating the opposite problem to the one you were trying to fix .....

The second method is loading the high cell/s. A 50W halogen bulb with a two wires soldered on to the bulb terminals and an alligator clip on each lead, Connect the alligator clips across the positive and negative of the high voltage cell, not the whole battery pack, this will drain some capacity out of that cell only, while the others are still charging .... basically a high resistance load that you can see and turn on/off as required ... as an example, the whole battery pack is charging at 5 amps, the light globe is drawing that 5 amps out of the high cell only, so the other cells can catch up.
If you have up to 4 cells in series, that are all high voltage, you can connect the 12V bulb across the negative on one end of the group, and the positive on the other end of the high voltage cell group. This will waste charging current out of the 4 cells in series, will the other cells are still getting charged.

Once you get the battery close to fully balanced at 100% SOC, all the cells above 3.5V, the permanently connected balancer will most likely be able to keep the cells balanced.

T1 Terry

 

[EVentropy]

It seems to me then, that the MG’s bms cuts the charge at the first cell reaching 3.5V and then starts balancing the low one(s) and calls that 100% as soon as it’s within “their” delta?
(From my first chart.) according to Everest’s graph. I guess it’s rediculous chasing <0.5%
Anyway. Sheesh, I think I will dial back the 3.6V HV cut off on boat LFP’s. They do start balancing with continued charging, at 3.4V though, but at minuscule current.
Cheers guys!
Apologies for thread steal…….