TL;DR #
A grouped bidirectional active balancing architecture using Buck-Boost circuits reduced SOC spread in a 12-cell retired LFP pack from 10.2% down to 2% during static equalization (98 min) and from 10.9% down to 1.94% during charge equalization (87 min) — with charge transfer efficiency reaching 70.06%. For buyers sourcing second-life battery packs or active BMS modules for stationary storage, this data establishes a concrete performance baseline: any supplier claiming effective cell balancing should be able to demonstrate post-equalization SOC spread below 2% under controlled test conditions. Before issuing an RFQ for BMS modules or second-life pack assemblies, require suppliers to submit equalization test data showing both static and dynamic performance against this threshold.
Overview #
If you’re evaluating active cell balancing for second-life lithium battery packs, the performance gap between what suppliers claim and what hardware actually delivers is significant — and the numbers here give you a reliable benchmark to hold them to. University-level research conducted on a 12-cell retired LFP pack sourced from a commercial electric vehicle application provides some of the clearest comparative data I’ve seen on grouped active balancing architectures. The test protocol used both static (resting) and dynamic (charging) equalization scenarios, with SOC estimated via square-root unscented Kalman filtering — a method that accounts for individual cell capacity variation, which matters enormously when working with retired cells whose usable capacity may have dropped to 60–80% of original nameplate.
The cells themselves came from a real fleet application: rated at 40Ah nominal, 3.2 V nominal, with a 3.6 V charge cutoff and 2.5 V discharge cutoff. After re-characterization, the 12 selected cells showed recalibrated capacities ranging from 29.8 Ah to 33.4 Ah — a spread that any passive balancing approach would struggle to handle efficiently.
For overseas buyers working with second-life cell packs or configuring multi-string stationary storage, understanding how cell consistency affects pack behavior is foundational before selecting a BMS topology. The cell balancing architecture — active vs. passive decision has direct cost and longevity implications that this dataset helps quantify.
Active Cell Balancing in Second-Life LFP Packs: What the Data Actually Shows #
The research tested a three-group architecture:12 cells split into three groups of four (P1, P2, P3), with intra-group balancing handled by a single-inductor centralized topology and inter-group balancing handled by a distributed inductor topology. This is not a novel concept — but the experimental rigor applied to retired cells, rather than new cells, is what makes this data procurement-relevant.
Intra-group circuit parameters: The balancing inductor was 0.35 mH, switching frequency 1 kHz, duty cycle 45%, maximum balancing current approximately 4 A. The inductor operated in discontinuous conduction mode (DCM) to prevent mutual interference between cells during energy transfer.
Inter-group circuit parameters: The distributed inductors between groups were 1.4 mH each. For m groups, the circuit requires2m−2 MOSFET switches and m−1 storage inductors — scaling linearly, which is a practical advantage for larger pack configurations.
The equalization strategy used SOC as the balancing variable rather than terminal voltage. This distinction matters: voltage-based balancing is faster to implement but fails on retired cells because capacity spread causes divergent SOC values even when terminal voltages appear similar. SOC-based balancing is more computationally intensive but ensures cells reach charge and discharge cutoff voltages simultaneously — which directly improves usable pack capacity.
Static equalization results:
| Parameter | Before Equalization | After Equalization |
|---|---|---|
| SOC spread (max−min) | 10.2% | 2.0% |
| Max deviation from group mean | 5.2% | 1.0% |
| Equalization time | — | 98 min |
| Charge transfer efficiency | — | 70.06% |
Charge equalization results:
| Parameter | Before Equalization | After Equalization |
|---|---|---|
| SOC spread (max−min) | 10.9% | 1.94% |
| Max deviation from group mean | 5.62% | 0.98% |
| Equalization time | — | 87 min |
| Charge transfer efficiency | — | 65.31% |
Charge equalization converged faster than static equalization (87 min vs. 98 min) because the charging current itself contributes to SOC redistribution — lower-capacity cells reach higher SOC states faster under constant-current charging, which partially assists the active balancing process.
Honestly, a 70% charge transfer efficiency is not impressive by itself — modern active balancing circuits in new-cell applications regularly hit 85–92%. But for retired cells with significant internal resistance variation (these cells had a nominal1 mΩ internal resistance at factory spec, with actual post-retirement values not uniform), 70% is a realistic field expectation, and suppliers claiming90%+ efficiency on second-life packs without supporting test data should be pressed on their methodology.
Cell Sorting Protocol and Capacity Recalibration for Retired Packs #
Before any balancing circuit can function correctly, the cells feeding it need to be properly screned. This part of the process is frequently underspecified in supplier datasheets, and it’s where second-life pack quality diverges most sharply between vendors.
The sorting protocol applied here is worth documenting for procurement purposes:
Step 1 — Visual screening: From a pool of 106 retired cells, 50 were selected based on physical condition.
Step 2 — Voltage screening: Open-circuit voltage (OCV) measured on all50. Cells with OCV between 2.4 V and 3.2 V (36 cells) were retained. Those36 were then charged at 10 A constant current for 2 hours; cells reaching above 3.0 V post-charge (30 cells) were advanced.
Step 3 — Capacity recalibration: Per national standard GB/T 34015–2017 (retired traction battery capacity test), cells were fully discharged, rested 15 minutes, charged at 1C to 3.6 V CC then CV to 0.05C cutoff, rested 30 minutes, then discharged at 1C to 2.5 V. This cycle was repeated three times; average discharge capacity across three cycles was taken as the new rated capacity.
Most retired cells in this dataset landed in the 24–32 Ah range. The 12 selected for balancing experiments showed recalibrated capacities between 29.8 Ah and 33.4 Ah — a 3.6 Ah spread on a 40 Ah nominal cell. That’s a 9% capacity range within a “matched” batch. This is the baseline you’re working with when sourcing second-life packs.
In supplier qualification, we saw a recurring pattern: three of six second-life pack suppliers sampled could not provide capacity recalibration data traceable to any standard test protocol. They quoted original factory capacity, not post-retirement measured capacity. This is a fundamental documentation gap that directly affects how you evaluate BMS sizing and expected pack runtime.
Most procurement teams don’t realize that GB/T 34015 — the Chinese national standard governing retired traction battery capacity testing — was updated specifically to address the capacity re-rating problem in second-life applications. If a Chinese supplier of second-life packs cannot reference this standard in their testing documentation, that’s a meaningful qualification signal.
The IEC 62619:2022 Safety requirements for secondary lithium cells and batteries does not specifically govern second-life cell re-rating procedures, but its safety requirements apply fully to any pack assembly — including those using retired cells. Buyers assembling second-life packs for stationary storage should also check compliance with IEC 61960-3 for secondary lithium cells in portable applications if the end product crosses into portable or UPS territory.
Balancing Strategy Selection: SOC-Based vs. Voltage-Based #
The choice of equalization variable has more procurement consequence than most buyers appreciate. Here’s the practical breakdown:
Voltage-based balancing is simpler to implement, lower cost in hardware, and faster to converge in new cells with consistent capacity. For a pack of new cells from the same batch, the OCV-to-SOC relationship is tight enough that voltage serves as a reasonable SOC proxy.
SOC-based balancing is computationally heavier — it requires an embedded SOC estimator (this implementation used a square-root unscented Kalman filter) — but it handles capacity-mismatched cells correctly. On a retired pack where individual cell capacities range±9%, voltage-based balancing would either over-balance high-capacity cells or under-balance low-capacity ones.
The threshold parameters used in this implementation:
- Intra-group balancing threshold (ΔSOC): 0.5%
- Inter-group adjacent difference threshold (ΔSOC_A): 0.5%
- Inter-group range threshold (ΔSOC_J): 1.0%
These thresholds represent a practical trade-off between balancing granularity and unnecessary switching activity. Setting thresholds too low wastes switching cycles on negligible differences; too high and you accept persistent SOC spread.
For stationary storage applications where SOC estimation accuracy directly affects usable capacity and cell longevity, specifying SOC-based balancing in your BMS procurement requirements is the right call — even though it adds cost. On retired cell packs, the alternative is leaving8–12% of pack capacity inaccessible due to premature cutoff from the weakest cells.
For reference on transport certification requirements applicable to any lithium pack assembly, UN 38.3 Recommendations on the Transport of Dangerous Goods — Lithium Battery Testing applies regardless of whether cells are new or second-life.
Practical Guidance for Buyers #
If you’re specifying a BMS for second-life LFP packs or evaluating suppliers of pre-assembled retired-cell modules, the data from this evaluation gives you specific, defensible thresholds. Don’t accept generic claims about “advanced balancing” — ask for equalization test results showing post-balance SOC spread, test duration, and charge transfer efficiency under both resting and charging conditions.
The numbers to anchor your specification: post-balance SOC spread ≤2%, achievable in under 100 minutes with a grouped Buck-Boost active balancing architecture on a 12-cell pack. Charge transfer efficiency in the 65–70% range is realistic for second-life cells; claims significantly above this without test data warrant scrutiny.
Cell sorting documentation is equally important. Any supplier providing second-life packs without a documented capacity recalibration procedure — using three-cycle average discharge capacity at defined C-rates — is selling you an uncharacterized product. The recalibrated capacity, not the original factory nameplate, is the value your system sizing depends on.
At CompactBES, our team connects global OEM buyers and energy storage integrators with verified Chinese manufacturers of BMS modules, second-life pack assemblies, and stationary storage systems — and we know which factories maintain rigorous cell sorting and requalification processes versus those that ship on appearance alone. If you’re configuring a sourcing inquiry for active BMS modules or second-life LFP assemblies, our team can help you navigate supplier capability quickly.
Need help identifying qualified suppliers for active balancing BMS modules or second-life LFP pack assemblies? Talk to our sourcing team →
Supplier Qualification Questions #
- Can you provide post-equalization SOC spread data for your retired cell packs, showing final SOC range ≤2% across all cells after active balancing under both static and charging conditions?
- What is the measured charge transfer efficiency of your active balancing circuit, and can you demonstrate≥65% efficiency with test logs from a12-cell or larger pack configuration?
- What cell sorting protocol do you apply to retired cells — specifically, do you perform a three-cycle capacity recalibration at1C discharge to 2.5 V and report the average discharge capacity as the new rated capacity?
- What are the switching frequency, duty cycle, and maximum balancing current parameters of your Buck-Boost balancing circuit, and how do these correspond to the inductor values (e.g., intra-group ≤0.5 mH, inter-group ≤1.5 mH) in your design?
- What SOC estimation method is embedded in your BMS — specifically, does it use a Kalman filter variant capable of handling per-cell capacity variation, and what is the estimated SOC accuracy (target≤1% error) under dynamic load conditions?
Sourcing Checklist #
- ☐ Supplier provides post-balance SOC spread test report showing ≤2% final SOC range across all cells in the pack
- ☐ Capacity recalibration documented per three-cycle average discharge at 1C to 2.5 V cutoff, with new rated capacity replacing original nameplate value
- ☐ Active balancing circuit achieves charge transfer efficiency ≥65% under charging conditions and≥70% under static conditions, verified with test logs
- ☐ OCV screening protocol documented — cells with OCV outside 2.4–3.2 V range are rejected before pack assembly
- ☐ BMS uses SOC-based balancing variable (not voltage-only) with intra-group threshold ≤0.5% and inter-group threshold ≤1.0%
- ☐ Balancing convergence time ≤100 minutes demonstrated for a≥12-cell string under controlled test conditions
- ☐ Pack assembly complies with IEC 62619:2022 safety requirements, with certification documentation available
Key Specifications Table #
| Parameter | Recommended Value | Verification Method |
|---|---|---|
| Post-equalization SOC spread | ≤2.0% | Static equalization test on full pack, SOC estimated via Kalman filter, record final min/max SOC across all cells |
| Charge transfer efficiency (active balancing) | ≥65% (charging), ≥70% (static) | Measure total charge transferred to target cells vs. total charge drawn from source cells during equalization cycle |
| Cell capacity recalibration range (retired LFP) | 24–34 Ah (from 40 Ah nominal) | Three-cycle 1C discharge to 2.5 V, average of three runs; reject cells below 24 Ah |
| Intra-group balancing inductor | 0.35 mH ± 10% | LCR meter measurement at switching frequency (1 kHz); confirm DCM operation |
| Inter-group balancing inductor | 1.4 mH ± 10% | LCR meter; verify energy transfer between adjacent groups at 45% duty cycle |
| Maximum balancing current | ~4 A (peak) | Current probe measurement during active equalization at specified duty cycle and switching frequency |
| OCV acceptance range (cell screening) | 2.4–3.2 V | OCV measured after ≥1 hour rest; reject cells outside this range before assembly |
Can’t find a supplier meeting these specs? Submit your requirements and we’ll match you within 48 hours.
References #
Data source: Grouped Bidirectional Active Equalization for Second-Life Lithium Battery Packs Using Buck-Boost Circuits with SOC-Based Balancing Strategy, T.-G. Song et al., Journal of the Electrochemical Society, 2025
Frequently Asked Questions #
Why is SOC chosen as the balancing variable instead of terminal voltage for retired cell packs?
Terminal voltage is a reliable SOC proxy only when all cells in a pack have similar capacity. In retired cell packs, individual cell capacities vary significantly — in the dataset here, from 29.8 Ah to 33.4 Ah on a 40 Ah nominal cell. Under these conditions, cells with lower capacity reach their voltage cutoff sooner, causing the pack to terminate early even though higher-capacity cells still have usable charge. SOC-based balancing accounts for per-cell capacity and ensures all cells approach their cutoff voltages simultaneously, which maximizes usable pack energy.
What is the difference between intra-group and inter-group balancing in a grouped architecture?
Intra-group balancing uses a centralized single-inductor topology to equalize any two cells within the same subgroup — it can address non-adjacent cells directly. Inter-group balancing uses a distributed topology where each adjacent pair of subgroups shares one storage inductor, enabling energy transfer between groups. The two levels operate sequentially: intra-group first until each group’s internal SOC spread falls below 0.5%, then inter-group to reduce spread across the full pack.
What charge transfer efficiency should I realistically expect from active balancing on retired cells?
The test data shows 70.06% under static conditions and 65.31% during charging — and these are honest numbers for retired cells. New-cell active balancing systems often cite 85–92% efficiency, but retired cells have higher and more variable internal resistance, which increases conduction losses during energy transfer. If a supplier quotes above 80% for a second-life pack without test evidence, ask for the measurement methodology and test conditions.
How long does active equalization take, and does it need to complete before the pack is used?
In this implementation, static equalization of 12 cells completed in 98 minutes; charge equalization completed in 87 minutes. For stationary storage applications, equalization typically runs during off-peak hours or at the start of each charge cycle. A well-designed BMS handles equalization concurrently with charging, which is why the 87-minute charge equalization time is more operationally relevant than the static figure for most real deployments.
Does using retired cells in stationary storage raise any certification concerns I should address before sourcing?
Yes. Even though the cells are second-life, any assembled pack must meet current safety standards for the target market. IEC 62619:2022 applies to secondary lithium cells and batteries industrial applications including stationary storage — it does not exempt retired or reconditioned cells. Additionally, if the pack will be shipped internationally, UN 38.3 transport certification requirements apply. Buyers targeting EU markets should also verify alignment with the EU Battery Regulation, which increasingly addresses second-life and recycled content documentation requirements.
Published by compactbess.com Technical Team | Request a sourcing quote
