Understanding SOC Drift

Why your battery percentage jumps - and how to fix it.

Below is the long version - what's actually happening inside the battery, why the percentage drifts in the first place, and exactly how the reset works.

The problem: the flat voltage curve

To understand SOC drift you need to know one thing about how lithium differs from lead-acid - the voltage curve.

The lead-acid "fuel gauge"

A traditional lead-acid battery behaves like a car's fuel tank. As you use energy, the voltage drops in a straight, predictable line. A monitor just looks at the voltage (say 12.2V) and knows almost exactly how full the battery is (around 50%). Easy.

Lead-acid voltage curve graph
Figure 1: Lead-acid voltage drops linearly, making it easy to read.

The lithium "cliff edge"

LiFePO4 doesn't behave like that. It's designed to deliver stable power for the entire usable range, so the voltage barely moves whether the battery is 80% full or 30% full. Then it falls off a cliff at the very end.

This is brilliant for your appliances - your fridge sees the same voltage all day - but it's a nightmare for the monitor. Voltage can't tell it the percentage, because the voltage isn't changing.

Lithium flat voltage curve graph
Figure 2: LiFePO4 maintains a flat voltage until the very end, making it hard to estimate remaining capacity.

How your monitor counts energy (Coulomb counting)

Since voltage is no help, the BMS uses a different method: Coulomb counting. External shunts (like the Victron BMV when paired with a TITAN battery) work the same way - and the TITAN battery range exposes the same data over Bluetooth so you can see it in the TITAN Lithium app on your phone.

Imagine standing at a door with a clicker, counting people entering and leaving a building:

  • You know the building started with 100 people (full).
  • You count 50 people leaving.
  • You assume there are 50 people still inside.

That's exactly what the BMS does. It measures the current (amps) flowing in and out over time, and from that works out how many amp-hours are left in the pack.

TITAN Lithium
Ah
BMS sensor



→ In (Charging)


← Out (Usage)

Figure 3: Animation showing the BMS counting energy as it enters and leaves.

Why the "jump" happens (the drift)

Coulomb counting is very accurate over a day or a week. But tiny measurement errors stack up over months.

  • A 0.1% measurement error doesn't matter today.
  • After three months of constant charging and discharging, that 0.1% error can have grown into a 10%, 20% or even 30% discrepancy.

Your monitor thinks the battery is at 40%, based on its running tally. But physically the cells are empty.

The "ghost load" effect

The TITAN BMS is deliberately calibrated to ignore extremely small currents - they can look like electrical noise. There's a good safety reason for this: if we made the BMS more sensitive (say 0.2A), random electrical noise could trick it into thinking current is flowing when it isn't. That matters because the BMS uses discharge current as one of its safety signals - it waits to see real load before it'll re-open the charge port after a full-charge cut. If noise mimicked discharge, the BMS could re-open the charge port on a full pack, leading to over-charge and a real fire risk.

The 0.6A threshold: the TITAN BMS effectively rounds anything below ~0.6A (about 8W at 12V) down to zero. So a single LED light, a USB phone charger or a TV standby indicator might be invisible to the monitor.

The result: a 0.4A load running overnight (10 hours) physically removes 4Ah of energy. The BMS, however, has logged 0Ah. Your screen says 100% when the battery is actually at about 96%. Multiply that across a few months and you get a noticeable drift.



Reality
(physical energy)








BMS screen
⚠


Ghost loads (<0.6A) sneak past the sensor… Real energy drains, but the screen stays at 40%… Voltage crash - BMS jumps to 0%.

Figure 4: How ghost loads create SOC drift and eventual jumps.

The fix: re-synchronising the system

The proper fix is to give the BMS a known reference point so it can reset its counter to a true 100%. We've engineered an explicit reset trigger into the protection logic for exactly this situation.

The solution: a full 100% charge

Charge the battery fully until one of two things happens:

  1. Pack over-voltage protection triggers (around 14.4V).
  2. Cell over-voltage protection triggers (around 3.60V on any single cell).

When the charger pushes the pack to either of those limits, the BMS briefly cuts the charge. At that exact moment, the BMS knows for certain the battery is physically full.

It immediately resets its internal counter to 100%, wiping out all the accumulated drift. From that point your monitor is back in sync with the actual energy in the cells.

BMS reset logic illustration
Figure 5: Reaching 14.4V forces the BMS to reset the SOC to 100%.

Making the charger actually get there

Here is the part that catches almost everyone. Knowing you need a full charge is one thing; getting your charger to deliver one is another, and most do not manage it on their factory settings.

We measured this across the fleet: of the batteries reporting usable data, roughly 87% never reach the voltage needed to trigger the reset. Not because anything is wrong with them, but because their charger stops fractionally short and nobody has ever had a reason to look. If your percentage drifts and you have dutifully charged to what the charger calls full, this is almost certainly why.

Two voltages, and they are not the same number

Conflating these is the single reason this fix fails for people, so it is worth being precise:

  • 14.4V is what the battery has to reach. That is a fact about the cells, not a setting you can change. Below it, no reset happens.
  • 14.6V is what to set the charger to. A charger set to exactly 14.4V only grazes the threshold, and once you account for voltage drop along the cables it may never quite arrive. Setting 14.6V gives it enough headroom to cross the line properly.

Set 14.4V and you will very likely repeat the whole exercise wondering why nothing happened.

What to do, whatever charger you have

  1. If the absorption voltage can be edited, set it to 14.6V, run one full charge, then put it back where it was afterwards. One charge is enough. This is the usual answer.
  2. If it cannot be edited, find out what it actually reaches. If the answer is below 14.4V, that charger will never be able to correct the reading on its own, and you will need to borrow or fit one that can. It is worth knowing either way.
  3. If you are on a Victron system with DVCC switched on, there is a second route: turn DVCC off for a single charge so the charger follows its own profile, then turn it back on. See the Victron integration guide for why DVCC causes this.

All of these are reversible, and none of them involves opening the battery or touching anything inside it.

Is 14.6V safe?

Yes, and it is a fair question to ask. Raising a charge voltage feels like the sort of thing that ought to come with a warning.

It does not, because the BMS is watching every cell individually the entire time and will protect them regardless of what the charger is asking for. That protection is what triggers the reset in the first place. The higher figure is not pushing the battery harder, it is giving the charger enough room to reach a limit the BMS was always going to enforce.

How to tell it worked

Open the Protections tab in the TITAN app and look at the cell overvoltage count. If the reset has happened, it will have gone up by one.

A counter labelled overvoltage going up sounds alarming, and it is not. It is simply the battery recording that a cell touched the top of its range, which is precisely the event you were trying to produce. Seeing that number increase is the confirmation you wanted, not a warning.