From Cell to Pack: What Actually Happens Inside a Battery Pack Factory
Updated: Sep 5

A buyer comparing two quotes for a 3.2 kWh LFP pack will often see the same cell manufacturer named on both, the same nominal voltage, the same capacity, the same connector. The quotes differ by 30%. The natural conclusion — that one supplier has better margins — is usually wrong. The difference is in steps the datasheet does not describe.
Incoming inspection and grading
Cells from the same batch are not identical. Capacity varies, internal resistance varies, self-discharge rate varies. A pack's behaviour over its life is set by its worst cell, so a pack built from ungraded cells inherits the widest spread in the box.
Serious pack manufacturers test incoming cells and sort them: capacity measured under a defined discharge, internal resistance measured by AC impedance, open-circuit voltage after a defined rest to catch high self-dischargers. Cells are then grouped so that every cell in a series string is closely matched. This step costs money and time. Skipping it is invisible on day one and expensive in year three, when cell delta has grown, balancing can no longer keep up, and usable capacity collapses.
Interconnection
How cells are joined to each other determines resistance, heat and mechanical reliability.
Nickel strip spot welding is common and adequate at modest currents, provided weld parameters are controlled and joints are pull-tested to a standard. Laser welding gives lower and more consistent resistance and is what high-current 2W and drone packs require. Ultrasonic wire bonding suits large formats and has the useful property of acting as a fuse on an individual cell.
The failure mode to design against is a joint that is good enough to pass final test and bad enough to fatigue after a year of vibration. This is why weld pull-testing on a sampling plan, and destructive testing of first articles, belong in the process rather than in the marketing.
Thermal and mechanical architecture
Cells need spacing for heat to leave, restraint so that they cannot move, and isolation so that a failure in one does not immediately become a failure in its neighbours. In LFP prismatic ESS modules this means compression fixtures with defined preload. In cylindrical 2W packs it means holders that maintain spacing and, increasingly, phase-change material or mica barriers between cells to slow propagation.
AIS-156 in its amended form requires a thermal propagation assessment: after a single-cell thermal runaway is deliberately initiated, the pack must give occupants a warning period before hazardous conditions reach them. That requirement is met or missed at this stage of design, not at the testing stage.
Enclosure and ingress protection
An IP67 rating is a claim about a validated design: gasket selection and groove geometry, breather vents that equalise pressure without admitting water, connector sealing, and a fastening pattern that maintains even compression on the gasket. It is verified by test, and the test report should be available on request. A pack described as "IP67" with no report behind it is describing an intention.
Electrical integration and end-of-line testing
The BMS is fitted, harnesses are routed and secured against chafing, and the assembled pack goes through end-of-line test. A good EOL sequence includes hi-pot testing for isolation, a capacity check against specification, an internal resistance measurement, a functional check of every protection threshold, communication verification, and a defined leak or ingress check.
Every pack gets a serial number and a record: which cell batch, which BMS firmware version, what the EOL test measured. That traceability is what allows a manufacturer to contain a problem to the forty packs that share a cell lot instead of recalling four thousand.
Formation, ageing and shipping
Packs rest at a controlled state of charge and are re-measured before dispatch, which catches high self-discharge assemblies that passed EOL. Lithium batteries then ship under UN 38.3 transport requirements at a defined state of charge — typically around 30% — in tested packaging.
What to ask on a factory visit
Ask to see the incoming cell grading data for a recent batch. Ask for a weld pull-test record. Ask what the EOL test measures and to see a report for a specific serial number. Ask how a cell lot is traced to a shipped pack. The answers separate manufacturers from assemblers more reliably than any brochure.




Comments