Why does the charge percentage (SoC) sometimes “jump”? (It also happens to lithium, especially LFP)

If you have a domestic battery connected to photovoltaics, you may see behaviour that is surprising:
the percentage of charge (SoC, State of Charge) does not always rise or fall in a perfectly continuous manner.
Sometimes it seems to make small “clicks” (e.g. 52% → 47% or 18% → 23%).

In the vast majority of cases, this does not indicate a failure: it is the natural effect of the fact that the SoC is a
intelligent estimation, not a direct measurement. And this estimate can update when conditions change
(power demand, temperature, charge/discharge phase, rest periods), or when the system recalibrates.

1) What is SoC really? An estimation, not a “sensor”

There is no sensor that “reads” the charge percentage like a float in the tank.
The SoC is estimated by BMS (Battery Management System) using measures such as:
current, voltage e temperature, together with a mathematical model.

In many architectures, a combination of:
coulomb counting (counting how much charge goes in/out over time) and
corrections based on voltage at rest (OCV) and on the temperature.
When conditions change or available information improves, the algorithm can “realign” the estimate,
and this may appear as a leap.

2) Because it is more noticeable at the beginning of life (or after reset/upgrade)

In the first days/weeks of use - or after a reset, a firmware update, or a major parameter change -
the system may go through a phase of calibration (often called self-learning).
At this stage, the algorithm refines the parameters with which it translates electrical signals into percentages.

When the algorithm finds new “reference points” (e.g. conditions close to end of charge or end of discharge,
or rest periods useful for reading the OCV), it can correct the estimate and make a snapshot appear.

3) Thermal excursion: because hot/cold makes the SoC more “nervous”.”

Temperature affects all batteries because it changes their electrochemical and electrical response.
In general, the temperature can change:

• internal resistance and voltage drops under load;
• the relationship between voltage at rest (OCV) e SoC;
• the dynamics of tension recovery when the load changes or the battery “relaxes”.

Result: for the same amount of energy actually present, the measured voltage may change differently depending on temperature and instantaneous power.
If the algorithm (also) uses voltage to correct the SoC, it can update the estimate when conditions change abruptly.

4) Does this also happen to lithium? Yes. And it is often more noticeable on LiFePO₄ (LFP)

This phenomenon is not “typical of sodium”: it is common to many chemistries (including lithium) because it depends on the way the SoC is estimated.

In particular, on batteries LiFePO₄ (LFP) SoC estimation is often more delicate because the curve
OCV-SoC has a very “flat” central area (a plateau):
In that range, the voltage changes little even though the percentage of charge changes a lot.
This reduces the voltage-based ’observability“ of the SoC and makes non-perfectly linear corrections more likely,
especially with noise, variable loads and low temperatures.

(For technical background: several works report that LFP has an OCV-SoC plateau that makes SoC estimation more difficult, and that small errors/voltage noise
can amplify into SoC errors; furthermore, the hysteresis between charge and discharge must be carefully managed).

5) Because, in some cases, sodium may be more sensitive to extremes (low SoC / high SoC)

Here it is important to be precise: there is no “sodium” as a unique behaviour, because it depends on the chemical specification (cathode/anode),
by the pack design and the BMS algorithm. That said, some experimental evidence on commercial cells shows that
in certain sodium-ion technologies the dependence of resistance/impedance on temperature e SoC
may be more pronounced than in LFP cells, and that in lower SoC regions, losses/inefficiencies may increase.

From a practical point of view, this means that close to the extremes (especially in low SoC and cold or high power):

• voltage may be more “sensitive” to load changes (higher voltage sag);
• the immediately usable energy can be temporarily reduced for voltage/power reasons;
• the SoC estimator can make more visible corrections because the electrical parameters change more quickly.

In addition, as with many chemistries, areas close to the limits may include regions with different tension gradients and phenomena
related to phase transitions or plateaus/slopes (depending on the materials): this can make voltage↔SoC mapping more “non-linear”
and therefore more prone to recalculations when temperature and load change.

6) How to read the image you insert below (and understand if it is “normal”)

When looking at a graph or screenshot with the SoC, to correctly interpret any “jumps”, always also check:

1. Battery temperature (if it changes a lot, it is normal for the estimate to be updated);
2. Charging/discharging power (sudden spikes change the voltage under load);
3. Phases of rest (at rest the voltage stabilises and the OCV is more informative);
4. Areas close to end of charge/discharge (where many systems apply conservative logic and corrections).

7) When to really worry

It makes sense to ask for verification if you notice:

• very wide and repeated shots (e.g. 20-30%) without changes in temperature or power;
• recurrent alarms or frequent sudden limitations;
• inconsistent behaviour between energy exchanged (kWh) and SoC variation over long periods.