SOC vs HV/LV delta and balancing….

[EVentropy]

Well, she’s finishing a history degree as a 68yro, “talked” me into buying a sailboat, that needed rebuilding, and then decided she hates sailing, so I’m not the only one!

M2CW, is that we have taken the signs (operators?) too literally and the measurement is just a delta. Lazy script writers choice or lack of , meaningful axis. Or it’s needed mathematically to define direction of change ie up or down to depict the instantaneous delta at any time . Yeah, whatever. Just leave off the (-) or (+) for that matter.
Voltage sag has got to be all over the place from high discharge to no discharge to full charge.

 

[T1 Terry]

If you take the most populated spot on the graph as the common cell voltage at low load, it makes a lot more sense. These guestimations are give or take half a brick or 0.05v

So the two short close cropped lines could be read as 3.31v , the peak 3.43v and low 3.24v.

My next guestimation is there are a few cells low voltage cells and only a single high voltage cell, the low drop is roughly half of the high peak over the average voltage.

T1 Terry

EDIT: I reread it after I posted it and I'm not sure that even makes much sense .... it does in my head but I doubt if I expressed it very well .... more coffee required maybe

 

[wattmatters]

Max - Min >= 0

 

[Coulomb]

Max - Min >= 0

I think Terry is saying that it's alternating (max - avg) and (min - avg), or something like that.

Edit: Perhaps not alternating, but each value is either max - avg or min - avg, whichever has the highest absolute value (the highest difference from average).

 

[T1 Terry]

I think Terry is saying that it's alternating (max - avg) and (min - avg), or something like that.

Edit: Perhaps not alternating, but each value is either max - avg or min - avg, whichever has the highest absolute value (the highest difference from average).

That's looking closer to what I meant to say The max over average is the high voltage/SOC cell on high regen, the negative below average voltage, under load, I'm thinking is shared across a few cells, not just one cell below the average voltage/SOC.
As can be seen by the difference between the highest voltage recorded and the lowest voltage recorded, it's not a very high voltage difference, but that doesn't mean it isn't a very high SOC difference, maybe even a few Ah difference that equates to a multiple of 3.? when it comes to Wh

T1 Terry

 

[EVentropy]

Ignoring the minus gobbledegook, what you are saying then is that the mean cell voltage is close to the min cell voltage at 98% SoC no load , so min cell could be any of a number of lower cell and there is only one runaway max cell voltage. Which is pretty much what I think for now. Which is borne out by bank voltage/104.
That aside:
I logged the return journey but switched off cell delta and logged max/min cell voltage (dual trace)

Now stretching the scale to get the time interval between data points in seconds it become a lot more understandable. As I have limited horizontal on phone I’ll go landscape to get more in. I’ll do two, one at beginning and one at massive cell V sag. (<3V)
I can’t seem to copy the file and share & also it needs to import into carscanner app to manipulate scales.

Last , below, is scale reduced to 20s interval at huge sag rhs of first (condensed trace).

None of this helps me wrt to how soc is measured, coulomb counting, mean cell V, tail current threshold, or what; of course without high cell V disconnect happening.
Charging finishes with the max cell hitting 3.66V on this bank. I find it difficult to believe both triggers are happening at same time and I would be surprised if the MG4 bms didn’t have this cell high cell facility. Probably written into ANCAP safety criteria lithium ion anyway. So far no success with bringing the bulk of cells closer to 3.45V or even a small pack mean voltage improvement at “100%”.
Next I will discharge to zero and do a full charge.
BTW, I watched Bjorn style2/2 live delta on the info centre and most of the -ve data was not happening. Occasional instant it went minus on extreme transitions. So some compilation problem.

 

[T1 Terry]

Be very careful deep discharging the battery with the low cell so low under load at 20:12, that is getting close to cell damage territory .... what SOC was the battery showing when that happened?

T1 Terry

 

[EVentropy]

~60%, bit scarey, but bearing in mind that was me flooring it and did return to 3.3Vish instantly load was off and regen was no longer happening.
Curious to know what is trigger for limp mode?
Min cell voltage or what? Also going to 0% I doubt will work for me as I have a steep 100m driveway to garage.
Cheers.

 

[T1 Terry]

60% SOC, that cell is a long way out of balance as far as SOC, but not rested voltage. Not unusual for LFP cells to bounce back like that, but that battery really needs all its cells up to 3.6v to saturation charge them ....... without access to the individual cell voltage monitoring wires, I don't know how you would achieve that ......
What is really needed is a graph with all 104 cell voltages, each identified by its own trace colour and line type ..... no idea how to do that without better access to the cells voltages either.

I'm going to need to build a workshed and put in a hoist it seems, so I can drop the pack out of the 51 to see how this can be done ..... so don't hold your breath, the minister for war and finance will have different views on how the $$ and time should be appropriated

T1 Terry

 

[wattmatters]

Min cell voltage or what? Also going to 0% I doubt will work for me as I have a steep 100m driveway to garage

Run the aircon?

 

[EVentropy]

Haha, I am familiar battles in the finances war room. I keep on leaving boat expenditure, well, at sea!
Yip, hear you on individual soc’s need to be equalised somehow. And the method I use on the boat isn’t doable.

Run the aircon?

On heat, right? So if I go to 5% on the road, 2.5 kWhr would take how long to dissipate with aircon?
I will watch minimum cell voltage carefully after 10%. Do you know the max power rating of B2L for stationary discharge?

 

[wattmatters]

On heat, right?

Depends on model location. In Australia MG4 does not use heat pump for heating so will draw more power.

 

[EVentropy]

Numbers that it would be nice to have if any one has the inclination to check for comparative purposes: MG4 LiFePO₄ 51kWhr:
1. Pack voltage as SOC goes 100% (just after)
Pack voltage after several hrs rest, no load.
2. Highest&lowest cell voltages as soc goes 100%.
3. Pack resistance.
4.Delta after 2hrs of no load.
Did I just hear someone muttering, “knock yourself out, the need to know wont’t last.”

I have ordered a V2L adaptor and will record behaviour of pack, 10–>0% and then
0% —>100% in terms of previous parameters.
I will post if any one is interested otherwise will just keep for my own log.
2023 MG4 Excite 51 Au/Nz 20k km

 

[T1 Terry]

Does anyone know how to record which cell is the high voltage and which cell is the low voltage on each sample taken?

T1 Terry

 

[wattmatters]

I have ordered a V2L adaptor and will record behaviour of pack, 10–>0% and then
0% —>100% in terms of previous parameters.

You can't discharge lower than 20% SOC with V2L.

 

[EVentropy]

Bugga…..
Does the heater allow 100% DOD?

 

[DixieTGS]

Does anyone know how to record which cell is the high voltage and which cell is the low voltage on each sample taken?

T1 Terry

Back around March or April carscanner reported some sensor values, many of which gave no reading or nonsense values even after having adjusted the sensor read timeout value. By guess work and looking what others had used for other MGs, I found a whole load of values which looked like battery voltages and temperatures, but didn't have any use for them and no body here seemed to know what to do with them.

I posted a couple of the dashboards I created:

OBD Data

@electricsidecar I appreciate your work to support the MG4. Having a good choice of solutions provides the best for everyone.

www.mgevs.com

The whole process of defining custom sensors and creating dashboards to put them on was very tedious, even using an editor on a PC to create them. Looking at the latest carscanner, the MG4 profile has changed a lot and has 24 battery min/max voltages listed in the sensors section, but no prebuilt dashboard to display them.

It should be much easier now to create dashboards like the examples I had in the link above, but using the pre-defined car scanner sensor definitions. I can't try anything out to see how useful the latest carscanner additions are as I'm away at the moment and the MG4 is at home.

A couple of comments though:

If all 24 battery values give valid readings for your car variant, then it still won't be a value per cell. So I'm guessing each reading could be for a group of at least 4 in series and maybe in parallel too?

Carscanner only recorded values to file if they were being displayed, and the largest dashboard can only display 30 values. This means you could log 24 Delta values (using a custom formula from min and max), but you couldn't log all 24 Delta values as well as the min or max which I suspect would be more useful.

 

[EVentropy]

If all 24 battery values give valid readings for your car variant, then it still won't be a value per cell.

So what the use of it? Begs the question of the delta trace off carscammer. Is it the mean delta off a module or are there separate circuits for each cell and it deducts previous V , for a per cell reading and no-one yet has sniffed out the PID? 104/24= 4.333 cells/mod. Configured unevenly to fit in space?

 

[DixieTGS]

So what the use of it?

I presume that if you had a single Delta value covering the whole of the battery and it told you there was a cell failure you'd have to replace the whole battery. If you had 24 deltas and one indicated a failure I would hope you would only have to replace 1/24th of the battery.

I don't know if it's any use at all regarding balancing.

At the moment we don't even know if the new values in car scanner actually produce good data. Some of the previous versions of car scanner have produced nonsense for some sensor values.

 

[EVentropy]

I presume that if you had a single Delta value covering the whole of the battery and it told you there was a cell failure you'd have to replace the whole battery. If you had 24 deltas and one indicated a failure I would hope you would only have to replace 1/24th of the battery.

Pretty sure the dealers/service agents are not going to be limited to max/min cell voltage for whole battery without cell ID, & which module.
Surely must be available thru the service portal programs and hardware interface that has been
uncovered.
Nissan leaf have 1 to 96 , each 2p, individual V readings in their graphical. The module configuration I don’t expect it’s for faulty cell ID & replacement reasons rather the distribution of ASIC’s amongst the modules for monitoring purposes.
AFAIK, they pull the the whole module and put a suitable specced replacement and the old is sent to be refurbished with errant cell to be replaced.
Minimum downtime. Anyway all cell voltages are available.