How to read a cycle-life claim

"1,000 cycles to 80%" is not a specification. It is the end of a sentence that began with five conditions, and without them the number can legitimately differ by a factor of three between two products using the same cell.

The five conditions

A claim missing any one of these five is not comparable with another claim that includes it.

The two conditions people most often leave out are depth of discharge and charge voltage, because both make the number look better. A cell cycled 100–50% at 4.15 V can post a cycle count several times higher than the same cell cycled 100–0% at 4.2 V — and both statements can be true at once.

Cycle ageing and calendar ageing are different clocks

A cell ages while it sits, not only while it works. Calendar ageing depends mainly on temperature and state of charge: a cell stored at high charge in a warm room loses capacity faster than one stored at partial charge in a cool one. A product that spends most of its life idle — backup power, seasonal equipment — is dominated by the calendar clock, and cycle-life figures tell you almost nothing about it.

• A pack that cycles twice a week will reach its warranty date long before its cycle count.
• A pack that cycles three times a day will be limited by cycle life, and by heat.
• Storage conditions belong in the warranty document as much as in the data sheet, because they are the difference between the two outcomes.

What a defensible test looks like

If you intend to publish a cycle-life claim, the test has to be described well enough for someone else to repeat it. That means stating the cell model and the batch, the charge and discharge rates, the voltage window, the temperature and how it was held, the rest period between cycles, and the retention threshold. A claim that lists all seven is verifiable; a claim that lists one is marketing.

For a product warranty, the useful approach is usually not to copy the cell manufacturer's number but to qualify the pack as built. The pack has different thermal behaviour, different charge limits and different depth of discharge from the test cell, so the pack figure will differ — and it is the pack figure the customer will experience.

Writing a warranty you can defend

1. Define the failure criterion in measurable terms: capacity below a stated percentage of the rated minimum, measured under a stated procedure.
2. State the operating envelope the warranty assumes: temperature range, charge rate, depth of discharge, storage conditions.
3. State what voids it: charging below 0 °C, water ingress, unauthorised BMS changes, cell replacement.
4. Publish the measurement method, so a dispute can be settled by measurement rather than opinion.
5. Keep the batch traceability that lets you identify which cells went into a returned pack.

The fourth item is the one most often skipped and the one that prevents most arguments. Without an agreed method, a customer measuring at 1 C and a supplier measuring at 0.2 C will disagree about the same pack indefinitely.

Qualifying your own pack, in four weeks

A full life test takes months. A qualification that tells you whether the design is in the right region takes weeks, and is usually enough to decide between two cell candidates.

1. Run three packs of each candidate at your real duty cycle, in your real enclosure, at the top of your ambient range.
2. Log cell surface temperature at the hottest location, pack current, and group voltages throughout.
3. Measure capacity at the start and at fixed intervals — weekly is enough for a four-week run.
4. Stop early if the temperature at maximum current exceeds the design limit, or if group spread exceeds the agreed band. Both are answers, and cheaper ones than a failed field return.
5. Fit a straight line to the retention points and extrapolate only far enough to rank the candidates. Do not publish the extrapolation as a cycle-life claim.

The distinction in the last item matters. Extrapolation is a design tool; a published claim needs a measured endpoint at the stated conditions. Using the first as the second is how a warranty becomes unenforceable.

Where claims go wrong in practice

• Copying the cell figure into a pack specification. The pack is a different thermal and electrical system; the number does not transfer.
• Quoting the best-case window. A figure measured 100–50% will not be met by a product that runs 100–10%.
• Ignoring the calendar. A five-year warranty on a cell with a good cycle count can still fail on calendar ageing at high storage charge.
• No stated measurement method. Turns every warranty discussion into an argument about method.

None of these is dishonest when it happens by accident, and all of them cost more to resolve later than to avoid at specification time.

Questions this guide does not answer in passing