SOX ALGORITHM
Embedded state estimation algorithms for every ME BMS. SoC, SoH, SoP, and SoE — tailored for both automotive and stationary applications, from NMC to LFP.
STATE ESTIMATION
Home › Software & Algorithms › SOX ALGORITHM
WHY IT MATTERS
EVERY DECISION STARTS WITH STATE
Core BMS functions depend on knowing battery state. Power limiting, thermal management, balancing, and range estimation all rely on accurate SoC, SoH, SoP, and SoE.
A 1 % improvement in SoC accuracy means more usable capacity, fewer unnecessary safety margins, and better performance throughout the battery's life.
SoX is the intelligence layer that makes every other BMS function work better.
SOX FEEDS EVERY BMS FUNCTION
SOX ALGORITHM SUITE
SoC · SoH · SoP · SoE
SOC
Power limits, balancing, fuel gauge display
SOH
Maintenance planning, warranty, capacity tracking
SOP
Drive/charge limits, grid services, performance
SOE
Range prediction, energy trading, dispatch
THE STACK
FOUR ALGORITHMS, ONE STACK
SoX covers the four fundamental state estimations that every BMS needs. Each algorithm is designed for both automotive and stationary operation, from NMC to LFP.
STATE OF CHARGE (SOC)
The fuel gauge. Hybrid estimation combines model-based observers with event-driven correction. Dynamically switches between open-loop and closed-loop estimation based on operating conditions. OCV curve recalibration in steep regions handles LFP's flat mid-range curve.
±1% NMC · ±2% LFP
STATE OF POWER
Predicts maximum charge and discharge power available at any moment. Considers temperature, SoC, age, and voltage limits. Configurable short-term and long-term prediction horizons for energy trading, grid services, and vehicle performance.
DYNAMIC POWER LIMITS
STATE OF HEALTH (SOH)
Tracks capacity fade and resistance increase over battery lifetime. Dual method blends SoC-based and voltage-based approaches for both accuracy and frequent updates. Rapid convergence after module swaps or significant operational changes.
SOHC + SOHR TRACKING
STATE OF ENERGY (SOE)
Estimates remaining energy in kWh — more accurate for performance prediction than SoC alone. Supports predictions at multiple C-rates (1C, 0.5C, 0.33C, 0.25C) for realistic dispatch planning.
±2% ACCURACY
APPLICATION COVERAGE
TAILORED FOR BOTH WORLDS
Most state estimation stacks optimize for one use case. ME's SoX algorithms handle both automotive and stationary profiles in a single stack.
AUTOMOTIVE
High SoC operation (50 % – 100 %)
Regular full charges between trips
Long rest periods (>2 hours) between driving cycles
NMC chemistry — monotonous OCV curve
STATIONARY
Narrow SoC windows or full 0%–100% range
High-frequency charge/discharge alternation
Rarely or never rested during operation
LFP chemistry — flat OCV curve (10%–90%)
ONE ALGORITHM STACK, OPTIMIZED FOR EVERY USE CASE.
Shape matching in low-SoC regions — where the OCV curve is steep and informative — ensures reliable recalibration even without rest periods. This is critical for stationary LFP systems that never fully rest. The hybrid estimation approach handles both automotive and stationary profiles without configuration changes.
CLOUD ENHANCEMENT
FROM EMBEDDED TO CLOUD
Battery-in-the-Cloud runs the same algorithms with significantly more compute — delivering higher accuracy, tighter operating margins, and continuous improvement.
The embedded SoX algorithms operate independently on every controller. Battery-in-the-Cloud adds a second estimation layer with more computational headroom, enabling tighter safety margins and more usable capacity.
Algorithm updates deploy to the fleet via OTA. Live SoX accuracy monitoring tracks estimation quality across every system in the fleet. Cloud enhancement is additive — the embedded algorithms always operate independently.
CLOUD STATE ESTIMATION
Higher accuracy than embedded-only, enabling tighter operating margins and more usable capacity.
OTA ALGORITHM UPDATES
Continuous improvement without site visits. Algorithm refinements deploy across the entire fleet.
ESTIMATION ARCHITECTURE
EMBEDDED SOX
Deterministic, runs on every controller
EIS RECALIBRATION
Impedance-derived ground truth, on-device
CLOUD SOX
Higher accuracy, more compute headroom
OTA UPDATES
Algorithm improvements pushed to fleet
PLATFORM COMPATABILITY
RUNS ON EVERY ME CONTROLLER
Same algorithm portfolio across all hardware variants. SoX ships as part of the SBS Software Platform.
SoX runs on every ME BMS controller variant — from the compact SBS Core to the stationary-optimized SBS Stationary. The same algorithm stack, the same accuracy, the same behavior.
COMPATIBLE PLATFORMS
SBS CORE
Compact controller for cost-optimized applications
SBS HIGH
Full-featured controller for demanding applications
SBS MULTISTRING
Multi-string controller for parallel battery systems
SBS STATIONARY
Stationary-optimized for BESS applications
| Parameter | Value |
|---|---|
| MCU | Automotive multicore processor (3 cores @ 300 MHz) |
| Contactor Drivers | 7 (6x normal + 1x PWM); 12V and 24V; 3.5 A peak, 1 A continuous |
| Pyro Driver | 1x with backup capacitor |
| Short-Circuit Detection | Up to 15,000 A (±275 A) |
| HV Measurements | 8 channels (2 pack + 6 generic) |
| HV Range | −1,000 to +1,000 V |
| HV Accuracy | ±1% (above 500 V), ±5 V (below 500 V) |
| Current Sensing | Integrated 35 µΩ dual shunt, ±1,000 A, ±0.5% accuracy |
| Peak Current | 500 A for 60 seconds |
| Isolation Monitoring | Passive, 0–100,000 Ohm/V, >5 MΩ |
| HVIL | 1x (12 mA constant current) |
| CAN | 3x CAN FD (up to 5 Mbit/s) |
| isoSPI | 1x (2 Mbps, bus topology) |
| Sensor Inputs | 3x 5V (no pull-up) + 3x 5V (10 kΩ pull-up) |
| Sensor Supply | 12V output (1 A) |
| Operating Temp | −40 to +85 °C |
| Power Supply | KL30 6–36V, KL30C 6–36V, KL15 |
| Sleep Current | <200 µA |
- Secure Boot
- Secure Update
- Secure Access
- HSM (ECC256 + SHA2)
- JTAG Lock
TECHNICAL SPECIFICATIONS
ALGORITHM SPECIFICATIONS
READY TO IMPROVE YOUR STATE ESTIMATION?
Get in touch with our team to discuss SoX integration for your battery system.