Prototype-level assessment of per-cell isolated dual-mode active balancing for low-voltage vehicle and auxiliary lithium battery packs

Cell imbalance reduces the usable capacity, efficiency, and service life of multi-cell lithium-based battery packs used in low-voltage vehicle and auxiliary energy-storage applications. This paper presents a prototype-level technical assessment of a dual-mode active balancing architecture based on galvanically isolated, per-cell low-power DC/DC modules. The proposed method combines scheduled charge-phase equalisation with discharge-time top-up of lagging cells, allowing energy to be redistributed from the pack bus to weaker cells during operation. The architecture, control logic, protection functions, and calibration procedure are described and evaluated on a 4-cell LiFePO4 bench pack assembled from Winston WB-LYP90AHA cells. In the available prototype test, under a 5 A constant-current discharge to 2.5 V per cell, discharge-time top-up increased the measured gross usable capacity from 84.1 Ah to 86.45 Ah, corresponding to a gain of 2.35 Ah or 2.8%. Because repeated experimental runs, direct module input/output energy logging, and continuous thermal traces were not available, this result is interpreted as a single-run proof-of-concept observation rather than as a statistically validated performance metric. To strengthen the interpretation of the available evidence, the study defines gross usable capacity, estimates balancing-energy overhead using a representative converter-efficiency range, discusses the sensitivity of the main control parameters, and benchmarks the evidence level of the proposed architecture against representative balancing approaches. The results indicate that per-cell isolated dual-mode balancing may provide a practical auxiliary route for improving gross usable-capacity utilisation in low-voltage lithium battery packs, particularly where modular integration, galvanic isolation, and compatibility with existing BMS interfaces are important. The manuscript should therefore be interpreted as a limited prototype-level technical assessment and proof-of-concept demonstration, not as a statistically validated automotive-scale BMS solution. Full statistical validation, dynamic load-profile testing, direct converter efficiency measurement, thermal characterisation, and vehicle-scale evaluation are identified as future work.