DaveinOlyWA said:
Another thought that has crossed my mind is Nissan instrumentation itself. Beginning to wonder how good it is and whether readings would be consistent across different vehicles of same trim, year, pack, etc?
This article says the analog front end for the Gen 2 Leaf is a Maxim IC.
https://www.maximintegrated.com/en/aboutus/newsroom/2018/nissan-integrates-maxim-battery-monitoring-ic.html
Digging on their website shows 4 publicly advertised BMS chips. Depending on which is used, the rated accuracy would be +/- 10mV (MAX17823B) or +/- 6mV (MAX17843) or +/- 4.5mV (MAX17852 and MAX17853) per cell across voltage and temperature. Or it could be a custom part that's not advertised, but I'd still expect similar accuracy.
That's just the analog front end. We don't know what sort of fuel gauging algorithm is used. Even with a perfect voltage measurement, it's not a simple thing to calculate SoC. The only two points that are well known are the 100% when end of charge occurs and 0% when the empty detection occurs. Any other point is an estimate based on predicting when the system expects full and empty detection will occur.
At mid-range, SoC is most difficult to determine with accuracy since you're far from the known points (coulomb counting drift has accumulated) and there's a pretty small slope in voltage vs. charge state (voltage based estimates are less accurate). I've seen my end-of-drive efficiency summary vary quite a bit depending on what battery range I'm operating in. My guess is that the SoC calculation error is showing.
But regardless of the error of the BMS system, it will put the car in turtle mode when it thinks the battery is low. So long as the BMS estimate of empty is conservative, the BMS estimate of empty will occur before true empty detection (this seems to be true for the Leaf, so long as no cells are "bad"). Meaning that the BMS estimate of empty might as well be considered true empty because it will limit vehicle function.