Cells off the same line are not the same cell
A battery pack is a group of cells asked to behave as one, and they never quite do. Cells leaving the same line differ in capacity, in internal resistance, and in how fast they lose charge at rest. Those differences are invisible in one cell and decisive in a string. Battery cell sorting and aging shrink that spread before assembly, and decide more of the warranty than any weld.
In a series string every cell carries the same current, so the smallest reaches both voltage limits first. Usable capacity is set by the weakest cell, not the average.
Three kinds of mismatch, and only one that balancing touches
Mismatch is usually discussed as one property. It is at least three. Capacity mismatch limits how much of the pack can be used, in both directions. Resistance mismatch behaves differently: under load a higher-resistance cell drops more voltage and runs warmer, because loss rises with the square of the current through it. Self-discharge mismatch is quieter, a leaky cell drawing charge from its parallel neighbours.
A battery management system helps with only the third. Balancing shifts charge but cannot create capacity or change how current divides between cells of different resistance. Treating the BMS as the answer is how a pack that tested perfectly on day one under-delivers in year three.
What the sorting station actually measures
Single-parameter sorting has obvious holes, which is why the robust workflow is staged: defect screening for damage, leakage and dimensional faults; formation and capacity grading; controlled aging with K-value screening; DCIR matching at a standardised state of charge; SOC normalisation; then grouping on capacity, DCIR, OCV and K-value together.
Two details separate a working station from a measurement bench. Measurement has to physically divert cells: where diversion is manual, the bin is chosen by whoever picks up the next cell, and the measurement achieves nothing. And identity has to survive to assembly. Automated systems sort up to about 6,000 cells per hour, and traceability is what makes the throughput worth having.
The aging room is a filter, not a waiting room
Aging is often read as queue time. It is a screening step with a defined thermal recipe. Makers run high-temperature primary aging, typically 45–60 °C for 24–72 hours, then room-temperature secondary aging at 20–25 °C for 48–120 hours. The first drives SEI stabilisation while amplifying small internal defects until they become detectable; the second suppresses OCV rebound before the final grading call.
What comes out is a filter. Cells whose voltage falls by more than roughly 5 mV across high-temperature aging are flagged suspect: a sustained drop is the signature of an internal micro-short. A sudden ACIR rise usually means a poor weld or insufficient electrolyte.
The cost is structural. Formation and aging together can represent up to about 30% of a cell plant's production capital expenditure and a quarter of its floor area, with process time of three days to three weeks. A line that treats the aging room as inventory rather than a filter pays for that capacity twice.
K-value: the number buyers should ask for
The most useful figure to come out of the aging room is K, the rate of open-circuit voltage decay, or ΔV/Δt across a fixed interval at controlled temperature. It has no universal pass mark; read it against the batch baseline. Ask for the test conditions and the distribution, not just a pass rate.
Published practice gives a sense of the windows. Cells are rested at least three hours before OCV is recorded, then accepted within about ±0.02 V spread for high-end work or ±0.03 V at the industrial end. DCIR is read at 50% SOC within about ±10%, and capacity is graded to roughly ±0.5–1%. A 30-day drop of 20 mV or less is the premium benchmark; 8 mV or less over 7 days is a working pass threshold.
| Parameter | Typical acceptance window | What it protects |
|---|---|---|
| OCV spread (≥3 h rest) | ±0.02 V high-end; ±0.03 V industrial | Voltage balance, BMS correction rate |
| DCIR spread at 50% SOC | About ±10% of batch | Heat rise and current sharing |
| Capacity spread in batch | Roughly ±0.5–1% | Usable capacity, set by weakest cell |
| Self-discharge / K-value | 30-day ≤20 mV; 7-day ≤8 mV | Micro-shorts and standby drift |
| Drop after high-temp aging | >5 mV treated as NG | Micro-short screening |
| Automated sorting rate | Up to ~6,000 cells/hour | Bin integrity at production speed |
What to write into the equipment spec
Sorting and aging equipment is bought on throughput and price. The clauses that decide whether it still works in year three are the ones about measurement conditions:
- The measurement set and the order of stages, not just the final bin.
- Acceptance windows as numbers, with the SOC and temperature each is measured at.
- The rest interval before OCV recording, and the interval over which K is calculated.
- How cells are physically diverted, and how bin identity reaches assembly.
- The aging recipe: temperature setpoints, dwell times and their tolerance.
- The NG rule, and who signs off on exceptions.
Then check both stations against the format and volume you plan in three years. NEGUP MACHINE builds the cell grading and sorting machine, the pack and module aging machine and the PACK/module EOL test system as one chain, so the grading decision and the aging record land in the same per-cell file.
On the floor
Sorting and aging are easiest to understand standing next to them. The clips below run from the battery cells loading section through the electronics build to the stations where cells are graded and settled.
Across the chain the discipline is the same: define the measurement conditions, keep identity attached to the part, and let the data, not the operator's judgement on the day, make the grading call. Our battery testing equipment and module and PACK assembly lines are specified that way.
WhatsApp +86 15307690902 | rain@negupmachine.com
Frequently Asked Questions
Do we need both sorting and aging, or can one replace the other?
They screen different faults, so one cannot replace the other. Aging exposes cells with abnormal voltage decay or latent micro-shorts that look normal immediately after formation. Sorting then measures and physically separates what remains. Skipping aging leaves self-discharge outliers in the batch; skipping sorting leaves the spread inside every bin.
How long does the aging stage add to our line?
Plan for 24 to 72 hours at elevated temperature followed by 48 to 120 hours at room temperature, which is why aging capacity is normally quoted in rack positions rather than in cells per hour. The dwell time is a process decision, but the rack count it implies is a capital decision you should make before you order the sorting machine.
What tolerances should we put in the purchase specification?
Write each acceptance window as a number and state the SOC, temperature and rest interval it is measured at, because the same cell will pass or fail depending on those conditions. OCV spread of about 0.02 to 0.03 V, DCIR within roughly 10% and capacity within 0.5 to 1% are common industrial starting points to adapt to your chemistry.
Can we retrofit sorting onto an existing aging room?
Usually yes, provided the aging rack can hand off cell identity along with the tray. The mechanical retrofit is the smaller part. The work is in binding the aging record to the cell ID so the sorting station grades on measured history rather than on a voltage read at the end of the line.
What does a quote need from us to be accurate?
Send the cell format and dimensions, the target throughput in cells per hour, the parameters you intend to grade on, and the tolerance windows you must guarantee to your own customers. With those four inputs we can size the sorting station, the rack count and the data layer, and return a station-by-station list with the measurement conditions spelled out.