State of Charge vs State of Health: Reading Battery Metrics
State of Charge vs State of Health: Reading Battery Metrics
Two numbers on every battery dashboard decide whether a pack is usable today and whether it will still be usable next year: state of charge (SoC) and state of health (SoH). Procurement managers routinely conflate them, and the confusion leads to stranded capacity, rejected warranty claims and packs written off years early. This guide separates the two metrics, explains how a real lithium battery manufacturer actually derives each one, and tells you what to demand from a supplier’s telemetry before you buy.
What SoC and SoH actually measure
State of charge is the fuel gauge: the charge available right now as a percentage of the pack’s current usable capacity. State of health is the odometer: the pack’s remaining capacity and internal resistance relative to when it was new, again as a percentage. SoC answers “can I run the load today?” SoH answers “how much life is left?” A pack at 30% SoC and 95% SoH is simply partially discharged; the same pack at 30% SoH is near end of life regardless of what the gauge says. Mixing these up is the single most common battery-specification error we see in RFQs.
How SoC is estimated
No cell directly reports charge. The BMS infers SoC from one or more signals: voltage under load, coulomb counting (integrating current in and out over time), and a cell model that corrects for temperature and aging. Voltage alone is a poor gauge near the middle of the curve, where a small voltage change spans a large SoC range; that is why cheap packs drift. Coulomb counting is accurate short-term but accumulates error unless it is reset by a full charge or discharge. The best systems fuse the methods and self-calibrate, which is exactly what you should ask a supplier to describe, not just claim.
How SoH is derived
SoH is harder because it requires a reference. A BMS establishes SoH by comparing delivered capacity in a recent full cycle against the rated capacity at birth, and by tracking the rise in internal resistance. Resistance grows as the cell ages, and capacity fades; both move the SoH percentage down. Because a true full cycle rarely happens in the field, mature firmware extrapolates from partial cycles using the cell model, then reconciles when a real full cycle occurs. Demand the method in writing: a vendor who only subtracts cycles from a rated count is not measuring SoH, it is counting.
Why buyers confuse the two
The confusion is structural. Both numbers are percentages, both appear on the same screen, and both decline over time, so a falling reading feels like “the battery is dying” whether it is discharge or degradation. The operational cost is real: a forklift operator sees 40% and parks a perfectly healthy pack, or a site manager trusts 90% SoH on a pack that was never calibrated and is actually at 60%. For a battery application solutions program spanning many packs, standardizing the SoH definition across the fleet is the fix, because inconsistent math across suppliers makes fleet comparisons meaningless.
Accuracy comparison by method
| Method | SoC error | SoH error | Best use |
|---|---|---|---|
| Voltage only | ±10–20% | n/a | Cheap consumer packs |
| Coulomb counting | ±3–5% short term | indirect | Motive power, EVs |
| Model-fused (EKF) | ±1–3% | ±3–5% | Fleet, medical, telecom |
| Periodic full-cycle recal | self-correcting | direct | Stationary banks |
The table shows why method matters more than the headline number. A pack quoting “±1% SoC” from a fused model is a different instrument from one quoting the same figure off voltage, and only the former gives you SoH you can defend in a warranty dispute. Insist on the method, not the marketing range, when you write the specification.
SoC accuracy in the field
In practice SoC drifts between calibrations. A pack that is never taken to a full charge loses its zero reference, and the gauge slowly lies. The remedy is operational: schedule a full charge or a controlled discharge every few weeks so the BMS can re-baseline, and confirm the BMS flags “SoC uncalibrated” rather than silently guessing. For a home energy storage system that trickle-charges daily, this usually happens naturally; for a motive pack topped up opportunistically between shifts, you must force it. The discipline is cheap and prevents the false “low battery” shutdowns that erode operator trust in the whole fleet.
SoH and the warranty conversation
SoH is the number your warranty actually rides on. A cycle-based or capacity-based warranty is only enforceable if both sides agree on how SoH is measured, at what depth-of-discharge and temperature, and against what end-of-life threshold (commonly 80% remaining capacity). If the supplier’s SoH math is opaque, the warranty is decorative. Put the SoH definition and the calibration method into the purchase order so the same formula judges a claim. A semi-solid-state battery carries the same requirement: new chemistry does not change the need for a defensible SoH baseline, it only shifts the model parameters.
Reading the BMS telemetry
SoC and SoH are only as good as the data behind them. A real telemetry export should give you per-cell voltage, pack current, temperature, cycle count, and a timestamped SoH estimate you can plot over time, not just a green light on a display. Pull the log monthly and watch the SoH slope: a pack losing two points a month is on a clear end-of-life path you can plan around, while a pack flat at 98% is doing its job. The buyers who win warranty and replacement battles are the ones whose logs show compliant use and a measured decline, not the ones waving a dashboard screenshot.
SoC and SoH across chemistries
The metrics behave differently by chemistry. LFP has a very flat voltage curve, which makes voltage-based SoC especially poor and pushes you toward coulomb counting or modeling. NMC’s steeper curve gives marginally better voltage indication but still needs fusion for accuracy. A sodium-ion battery is younger, so its SoH model database is thinner and you should demand cycle data at your duty to validate the supplier’s estimate. Regardless of chemistry, the procurement rule is identical: require the method, require the calibration, require the export, and you can compare packs on equal terms.
A procurement checklist
- Require the SoC estimation method in the datasheet, not just an accuracy figure.
- Require a written SoH definition tied to capacity and resistance at a stated DoD and temperature.
- Require periodic recalibration behavior (full-cycle re-baseline) to be described.
- Require a telemetry export format you control, not locked in a vendor cloud.
- Put the SoH method into the purchase order to make the warranty enforceable.
- Standardize SoH math across suppliers so fleet comparisons are honest.
Common mistakes to avoid
The usual failures are predictable. Buyers accept “SoC ±5%” without asking the method and discover voltage-only estimation in the field. They treat a falling SoH as a fault instead of a trend and replace a pack with years of life left. They compare SoH percentages from two suppliers using different definitions and rank the pack that is actually worse as better. And they never log telemetry, so when a cell fails early they cannot prove the pack was used within rated conditions. Each mistake is avoidable with the same three demands: method, calibration, export.
SoC and SoH in fleet operations
At fleet scale the two metrics become management instruments. A dashboard that rolls up SoH across hundreds of packs lets you schedule replacement before capacity cliffs strand a vehicle, and lets you prove to finance that a pack retired at 80% SoH still has resale value as a second-life unit. SoC roll-up tells dispatchers which units are ready for a long shift and which need a charge, but only if every pack reports on the same method. Mixing voltage-only and model-fused estimates produces a fleet average that is wrong in both directions. Standardize the BMS reporting contract across the whole fleet, and the SoC/SoH data becomes something you can act on rather than a collection of incompatible gauges. The operational savings from planned replacement alone usually exceed the cost of the telemetry that produced it, and the same data defends every warranty claim you file.
SoH and second-life valuation
SoH is also the number that sets a pack’s residual value. A unit at 80% SoH with a documented cycle and temperature history is a credible second-life asset for a solar buffer or low-rate stationary use; the same pack with no history is a mystery brick nobody will buy. Buyers who log throughput and temperature from day one can sell the pack at end of rated life instead of scrapping it, and that residual value should appear as a credit in the original total-cost model. Require the supplier to export state-of-health in a format you own, not locked in a vendor cloud, so the pack has a buyer when you are done with it. This also future-proofs you against tightening battery-waste regulation, turning a disposal cost into residual value and strengthening your ESG position with a number auditors can verify rather than a slogan they must trust.
Frequently asked questions
Is SoC the same as SoH? No—SoC is the current fuel gauge, SoH is the remaining life percentage relative to new; they decline for different reasons.
How accurate is SoC? Voltage-only is ±10–20%; a fused model with recalibration holds ±1–3%, which is what fleet and medical use need.
Can SoH be measured directly? Only by a real full charge-discharge; in the field it is modeled and reconciled, so demand the method and calibration.
Why does SoH matter for warranty? A capacity-based warranty is only enforceable if both sides agree on the SoH definition and how it is measured.
Written by Karl at China Battery Technology. Send us your telemetry spec and we will set the SoC/SoH method—request a quote for a measurable BMS definition.
Related Reading
- battery application solutions hub
- Prismatic vs Cylindrical vs Pouch Cells: A Buyer’s Comparison
- Lithium Battery Nominal Voltage: Cells, Modules and Packs
