What a Cylindrical Cell Teardown Shows That Grading Data Cannot

A pilot batch of 21700 cells comes out of formation looking healthy. At grading, three cells from the same rack position run 3-4% below the batch mean on capacity, and their DCIR sits just above the upper spec. The formation recipe has not moved, the coating logs are flat, and the electrolyte lot matches the previous batch. The team has three theories: coating weight drift, incomplete electrolyte wetting, or a winding tension excursion on one shift. Testing all three with a full DOE would burn a week of rack time and a roll of electrode stock. Opening the three cells takes twenty minutes and ends the debate.

This is the normal rhythm of cylindrical pilot work. Grading data tells you a cell failed; it rarely tells you where. Teardown, or controlled cell disassembly and inspection, is the step that connects the two.

Grading Gives You the Symptom, Not the Location

Capacity, DCIR, OCV, self-discharge: every number from the grading machine is an integral over the whole cell. A 3% capacity shortfall could come from cathode loading, anode balance, electrolyte starvation, separator wetting, or a resistive tab weld. Electrical tests cannot separate those causes, and the longer the suspect list, the more tempting it is to run a multi-factor experiment. For a single anomalous channel in a pilot batch, teardown is usually cheaper and faster than the experiment.

The discipline that makes teardown worth anything: keep each cell's grading record attached to it. Pilot-line grading machines log per-cell values by channel, and cells sit in known rack positions, so you know exactly which numbers belong to the cell on the bench. You are correlating a measured quantity with a physical state, and that correlation is what turns a teardown from anecdote into evidence. If you open cells without their grading records, you are guessing.

We wrote earlier about why cell grading becomes the bottleneck in pilot cell qualification. The teardown is the natural next step after grading flags a cell: the tester decides that a cell is bad, the disassembler decides why.

Open in a Fixed Order, and Photograph Everything

The order matters because each step can destroy the evidence for the next one. A sequence that works well for cylindrical cells:

1. Jelly roll position, before anything else. After decapping, check the roll's position relative to the can bottom and the top insulating washer before you unwind anything. A shifted roll explains shorting, capacity loss, and DCIR rise all at once.
2. Tab welds next. Inspect both weld interfaces, tab to can bottom and tab to terminal, for partial welds, burn-through, or missing weld spots. A weld that passes the pull test can still be resistive; the failure mode shows up here long before it shows up in cycling.
3. The unwinding. Unroll slowly and look at edge alignment and telescoping. This is where winding tension problems leave physical evidence, and it is why tension control is a common topic in cylindrical pilot cell qualification.
4. Electrolyte distribution. Note wet versus dry regions across the roll, especially at the core. A dry core after weeks of rest points at filling or wetting, not at chemistry.
5. Foil and separator condition. Discoloration, pitting, or crystalline deposits on the copper foil tell a moisture or electrolyte-decomposition story that no electrical test will show you.

Photograph each stage before you move the part. The photo is what the process engineer will actually argue over.

Decrimping Without Destroying the Evidence

The hard part of cylindrical teardown is opening the can without wrecking the roll. Prying the cap off with pliers distorts the can wall and tears the outer windings, and what you learn from a mangled roll is mostly that you mangled it.

A rotary disassembler does this cleanly: the cell spins in a die while a tungsten carbide cutter follows the crimp, separating cap from can without cutting into the roll. In the pilot line configurations Wangsheng supplies, the blade runs at 1,000 rpm against a die rotating at 300 rpm, and the machine covers the common small-format sizes, 18650, 21700, 26650, and the smaller 10-series cells. Two practical rules on top of the tool: do the teardown in a dry box or dry room, because a wound cell starts reacting with air the moment you open it, and cut slowly near the crimp, because the blade should follow the groove, not force through it.

Reading the Findings: Finding to Process Step

Once the roll is out, map what you see back to a process step. The table below is the mapping that comes up most often on cylindrical pilot lines:

Notice what the table does not contain: chemistry. In our experience, most pilot-cell anomalies traced by teardown end at mechanics, alignment, tension, weld energy, wetting time. That is a useful bias to have. It stops you from changing the formation recipe when the real fault is a loose mandrel.

How Many Cells to Open, and Which Ones

There is a trade-off between evidence and inventory: every cell you open is a cell you cannot cycle further or ship. On a pilot line that cost is low and the information value is high, so the bias should be toward opening.

The sampling rule we recommend: when one grading channel fails, open two failing cells from that channel plus one healthy neighbor as a control. When failures spread across channels, open cells from the extremes, worst and best, because the spread itself is diagnostic. Two or three cells per condition is usually enough to see a pattern; beyond that you are collecting anecdotes, not data.

The calculation flips on a production line. There, teardown is reserved for first articles, major changeovers, and customer complaints, because the throughput cost of pulling good cells is real.

When Not to Tear Down

Teardown is not the answer to every anomaly.

• If a single isolated cell fails with no pattern, check the test fixture first. A loose contact on the grading channel mimics a bad cell, and no amount of disassembly will find a connection problem.
• If you suspect contamination, teardown shows that something is wrong but not what it is. Send electrode or electrolyte samples for SEM/EDX or ICP analysis instead of guessing from the bench.
• If you need reliability-level failure statistics, one teardown is not a distribution. You need a defined sampling plan and a log kept over batches.

And when you buy the disassembler, ask the supplier three questions: does it cut the crimp or crush it, does it need tooling changes between your can sizes, and how long does a full teardown take. The answers decide whether the tool lives next to the grading machine or stays in a drawer.

A Teardown Log Compounds

The first teardown is a fix. The tenth is a database. If every opened cell gets a photo set, a measurement sheet, and its grading record attached, then after a few batches you can answer questions like "did the anode lot change shift the failure signature?" from the log instead of from memory. That is the quiet way teardown pays for itself on a cylindrical pilot line.

When you spec a cylindrical battery pilot line, put the disassembler next to the grading machine and make teardown a standard step in qualification. It is the fastest root-cause tool most pilot labs are missing. If you are setting up that line and want the full equipment scope, from winding through grading, contact Wangsheng and we will walk through the configuration with you.