SOC vs HV/LV delta and balancing….

[gwatuk]

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.

I charge our MG4 SE SR with a granny charger. The mains supply to the charger is switched on with a Shelly relay and this also monitors the mains power. And I can plot this with Home Assistant automation.
Not sure if this helps but I've just finished a balancing charge and this is what the mains power looks like:

 

[EVentropy]

That just looks like the switching of the contactor from full power to no power and resumes for a short period then stops charge at 7:32. Do you know what max cell voltage was at
beginning of cycling and at 7:32?
That would explain the contactor at rear “clacking” at ~99% on mine.
I suspect my last experience of not reaching 100% before stop charge was that max cell V threshold was reached first. (3.66V) This varies charge to charge.
Whether any input power fluctuations would show from switches of balance resistors I doubt as all it dissipating power with heat.(AFAIK from T1 Terry)
My min V cell does not ever increase V from this procedure, in fact it decay about 150mV and max cell reduces very slowly and ceases at 150mV to decay even slower.

 

[wattmatters]

Balance charging at end of a charge session draws around 50-100 W:

 

[EVentropy]

To do what? (No pun)
1. Lift the V of minimum cell?
2.Cool the resistor(s) taking the cell(s) with high V? To lower V.

As I stated my minimum cell V continues to settle while “balance” is supposedly happening. So #1 is not correct.
Don’t know how/if any cooling is taking place.
I haven’t found any explanation (non-speculative) for the “balancing facility “ on the MG4 sr that can be backed up with cellular voltage increase decrease.
My diy boat install has bms graphical representation of cells that are being balanced that can be corroborated with a multimeter. You can see exactly what is happening.
Oh, I’ll repeat the mantra before anybody else does. Keep your LiFePO4 at 100% as short a time as possible.
Edit: Also, as Terry T1 said, unless it is saturated (held at 3.65V for decent time) surface charge will blow off by itself. The rate I cannot speculate on as haven’t needed to know. Saturated, afaik is when tail current stabilises at <o.xA . Andy (OGG) goes to zero, . But that could be rounding down? I’m not brave enough and who cares for <0.5%. Sorry for drift away from MG4 balancing.
I suspect MG have 100% soc criteria shy of knee anyway. So no need to be paranoid about it.

 

[EVentropy]

Also, as suggested: discharge to 98% and recharge to 100% (lowest of displayed which is 99.xx on mine) twice, revealed no change in delta or battery mean voltage rested. (Delta <10mV , no load. SOC 98%, ~3.4V)
Next trial will be to 0% to 99.xx%.

 

[T1 Terry]

Also, as suggested: discharge to 98% and recharge to 100% (lowest of displayed which is 99.xx on mine) twice, revealed no change in delta or battery mean voltage rested. (Delta <10mV , no load. SOC 98%, ~3.4V)
Next trial will be to 0% to 99.xx%.

That means the BMS doesn't attempt to saturation charge the cells and knows it by dropping 2% off the SOC, it hits 100% but drops back to 98% at rest. The 0.01v is a good balance if that is at the end of charge 100% and not rested voltage, if they were to all be at 3.45v or 3.5v and that close together, you would have a true 100% SOC, but still, that is fairly close. Did you try the headlight load to bring the DC to DC online to load the battery and see what the delta was then?

T1 Terry

 

[EVentropy]

That means the BMS doesn't attempt to saturation charge the cells and knows it by dropping 2% off the SOC, it hits 100% but drops back to 98% at rest. The 0.01v is a good balance if that is at the end of charge 100% and not rested voltage.

Unfortunately no. I wish, but don’t think so. The delta drops from 0.290V to ~0.093V over 2hrs no load after charging stops.By then SOC is 99.3%.

The 0.290V delta is at max cell V of 3.66V and I suspect max Cell limit is reached and bms disconnects charge bus before 100% soc algorithm can and then switches to balance as programmed. As you know 3.65V is industry standard for this bms feature, and a very prudent one at that.(for LiFe) To get delta sub 10mV no load I have to drive it and guesstimate as the delta varies with load. Somewhere 98-99% it get there.

Did you try the headlight load to bring the DC to DC online to load the battery and see what the delta was then?

T1 Terry

No, but that is a good idea to try and bring the pack up, while keeping the runaway under 3.65V.
It would be better if I could get at the pack, isolate the cell and discharge back to pack voltage, Rince and repeat (Everest style).
If I get no satisfaction from MG tech, I might give it a go. But car is only 15 months old with 20k km, so a bit on the nose really. Besides I ruptured my Achilles a week ago so not ideal for crawling around awkwardly amongst HV.
Guess that would doom my warranty?
Car scan after 1.5hrs

 

[Coulomb]

No wonder these XPowers are so quick:
In the above screen dump, I see:

Kinetic energy: E = ½mc²

That's a nuclear formula converting mass completely into pure energy! (c is the speed of light, not the speed of the vehicle). Well, it goes half way, Einstein's famous equation doesn't have the ½ factor.
/geek-mode

 

[EVentropy]

Is it some kind of puzzle?
Rotate c ninety degree anticlockwise.
At least potential energy Is zero at sea level as I don’t recall my vehicle being submersible in the manual at zero percent.

 

[Everest]

1. Lift the V of minimum cell?

I do that with my home batteries. But not sure it's so easy to plug in your bench PSU to the HV pack of an MG4

 

[EVentropy]

What I was thinking was pull the master HV fuse, get to the cell terminals (possibly extremely tedious) to identify runaway cell, connect suitably specced load (12V bulbs or some such) and drop cell voltage to pack cell mean voltage or slightly lower (say 3.3V). Remove, then replace fuse and charge with MG 16A or 32A regular charger.
Repeat until you get pack voltage at knee or where you want it, as Terry suggested. This would stop the bms max cell voltage cutoff preceding 100% SOC cutoff programming.
I played around with this on boat LiFePO4 CALB 230Ahr cells briefly and it worked.

 

[Everest]

What I was thinking was pull the master HV fuse, get to the cell terminals (possibly extremely tedious) to identify runaway cell, connect suitably specced load (12V bulbs or some such) and drop cell voltage to pack cell mean voltage or slightly lower (say 3.3V). Remove, then replace fuse and charge with MG 16A or 32A regular charger.
Repeat until you get pack voltage closer to knee as Terry suggested. This would stop the bms max cell voltage cutoff preceding 100% SOC cutoff programming.
I played around with this on boat LiFePO4 CALB
230Ahr cells briefly and it worked.

once a year I like to spend an afternoon bleeding off voltage from my high cells with a 12V headlight bulb and charge up the laggers with a bench PSU set to 3.5V.

Fine for half a dozen EVE 280Ah's in my 16s pack, but not much fun underneath an MG with 104 of them

 

[EVentropy]

Well done! Awesome, and I only have 8!

Not contemplating reconfiguring 104S to 104P to balance. My other half already thinks I have a screw loose. Also My 10A PSU would have a coronary.

 

[EVentropy]

For those that are curious: Graph of cell delta vs time of my pack after charging- driving.
Note:
1. Start at 83ish mV 4hrs after charging stopped.
99% SOC.
2. Start driving flat road to base of hill and drive conservatively with mean delta decreasing.
3. Stop at 500’ vertical and get a low load/stable load vs time. Short block of stablish horizontal reading at <10mV. ~98% SOC (a few minutes)
4. proceed to 1000’ vert. And on to the city using data gathering as an excuse for bursts of extreme acceleration and various styles of driving. Last section is back on flat in the city.
I tried superimposing SOC but the gnome that does it wasn’t home.

Interested in the various responses to + & - voltage variations. I guess if increased load caused increased delta owing to voltage sag on the reactive cell(s), regen does the opposite?

 

[T1 Terry]

For those that are curious: Graph of cell delta vs time of my pack after charging- driving.
Note:
1. Start at 83ish mV 4hrs after charging stopped.
99% SOC.
2. Start driving flat road to base of hill and drive conservatively with mean delta decreasing.
3. Stop at 500’ vertical and get a low load/stable load vs time. Short block of stablish horizontal reading at <10mV. ~98% SOC (a few minutes)
4. proceed to 1000’ vert. And on to the city using data gathering as an excuse for bursts of extreme acceleration and various styles of driving. Last section is back on flat in the city.
I tried superimposing SOC but the gnome that does it wasn’t home.

Interested in the various responses to + & - voltage variations. I guess if increased load caused increased delta owing to voltage sag on the reactive cell(s), regen does the opposite?

View attachment 33531

This where a graph of each cell voltage would be the go, then you could see which cells are giving you grief and if it is the same cell, indicating high resistance, or one high cell and one low cell.

The graph doesn't make a lot of sense, how do you have minus mV delta ....

T1 Terry

 

[EVentropy]

Lack of Kalman filtering, statistical noise, the way the delta is measured, back emf, who knows.
See it seems to just lag or precede a similar amplitude in opposite directions. I’ll do it again with a regime of fixed acceleration deceleration to time.
“This where a graph of each cell voltage would be the go, then you could see which cells are giving you grief and if it is the same cell, indicating high resistance, or one high cell and one low cell.”
Absolutely. I Put a deposit on a 2024 leaf because of that ability and pulled it because of NMC and no cooling, in favour of MG. I see someone posted pid’s for cell voltages of MG4 so only matter of time before someone scripts it into a graphical in real time.
Thing is, you still have to pull the lid on the bank and replace the he cells that’s out of whack.
So perhaps just knowing there is a delta over a certain threshold is enough.
I am not very optimistic about learning much from the MG technicians here as opposed Leaf technicians in Christchurch. Understandable as they (Nissan leaf) have had a10yr head start.
Man, they are right up with the hardware nuances, and vocal about it.
I used to love my Nissan patrol. Then there was the 1964 Morris Oxford woodie van.

 

[T1 Terry]

Lack of Kalman filtering, statistical noise, the way the delta is measured, back emf, who knows.
See it seems to just lag or precede a similar amplitude in opposite directions. I’ll do it again with a regime of fixed acceleration deceleration to time.
“This where a graph of each cell voltage would be the go, then you could see which cells are giving you grief and if it is the same cell, indicating high resistance, or one high cell and one low cell.”
Absolutely. I Put a deposit on a 2024 leaf because of that ability and pulled it because of NMC and no cooling, in favour of MG. I see someone posted pid’s for cell voltages of MG4 so only matter of time before someone scripts it into a graphical in real time.
Thing is, you still have to pull the lid on the bank and replace the he cells that’s out of whack.
So perhaps just knowing there is a delta over a certain threshold is enough.
I am not very optimistic about learning much from the MG technicians here as opposed Leaf technicians in Christchurch. Understandable as they (Nissan leaf) have had a10yr head start.
Man, they are right up with the hardware nuances, and vocal about it.
I used to love my Nissan patrol. Then there was the 1964 Morris Oxford woodie van.

Now, there is a great contender for an EV conversion.
I wonder if the LFP battery could be reconfigured to fit through the rear floor, maybe 50/50 to use the space under the vehicle the tailshaft used, the transmission hump for the electrics requiring that was able to be accessed from the top or the complete module removed from underneath, then the front for the remained of the cells where the ICE once was.
The rear motor replaces the diff, maybe even the whole rear module from the MG would fit .... maybe .... I seem to remember the Morris Oxford was still a tad on the narrow side .....

T1 Terry

 

[EVentropy]

That would be dream project! Certification would be a frickn nightmare. How do you find engineers whose design credentials would satisfy Vin ppl for that type conversion?
How are your projects going? Do you have a website/link for them.

 

[T1 Terry]

That would be dream project! Certification would be a frickn nightmare. How do you find engineers whose design credentials would satisfy Vin ppl for that type conversion?
How are your projects going? Do you have a website/link for them.

Even if I did have a website, the write up would look like I did it in a snow blizzard I've fixed the Ford 6.8 ltr V10 in the 30ft Winnie so it runs on all 10 cyls again, the torque and acceleration are a tad intoxicating so the wallet suffers at the LPG pump ..... it gets out and hides if I stop with the petrol filler near a pump Over AUD$500 to fill up the petrol tank at the current Christmas/New Yr fuel prices ..... so it runs on LPG and only half that to fill up.

I have the line up of donor vehicles and a collection of vehicles that would love to be brought into the 21st century, just need a shed and the weather to cool down a ted to get started, the high 30's and low 40's in the shade, is just a little too much without a shed to keep the sun off, everything ends up too hot to touch ......

T1 Terry

 

[Everest]

Not contemplating reconfiguring 104S to 104P to balance. My other half already thinks I have a screw loose. Also My 10A PSU would have a coronary.

LOL! You can just charge up one cell with an isolated bench PSU whilst all the others are still connected. I'd only do that on my home battery though where I can only zap myself with max of 48Vdc.

Interested in the various responses to + & - voltage variations.

As @T1 Terry already said, a 'negative' delta makes no sense at all. You can't have a negative delta, it's just a measure of the difference between highest and lowest cell.