What Actually Determines Li-Ion Cycle Life

An engineer pulls up two cell datasheets during pack selection. One says 600 cycles. The other says 2,000 cycles. Nothing on either page says what "cycle" means: no depth of discharge, no C-rate, no capacity threshold, no test temperature. There is no way to tell if those two numbers describe the same thing, and no way to know which cell will actually outlast the other in a real system.

That gap matters more than it looks. Size a pack against the wrong cycle-life assumption and you get one of two outcomes: an oversized, overpriced pack built to a number nobody needed, or a fleet that needs mid-program replacement because the real number was lower than the datasheet implied. Both are program-level failures. Neither is a footnote.

In short: li-ion cycle life is not one number. It's the output of a specific test, depth of discharge, C-rate, capacity-retention threshold, and temperature, applied to a chemistry, and changing any one of those conditions changes the reported number even for the identical cell. Comparing two datasheets means normalizing for all four variables first. Applying a rated cycle count to your own platform means converting your real mission profile into equivalent full cycles, not counting charge events. This post covers both, plus the pack and BMS-level design choices that actually protect the number once it's built into a real pack.

Why Cycle-Life Numbers Aren't Comparable Across Datasheets

A cycle-life figure is a test result, not a physical constant. Four variables determine it: the capacity-retention threshold that defines "end of life," the depth of discharge (DoD) per cycle, the charge and discharge C-rate, and the test temperature. A datasheet that states a cycle number without stating all four hasn't given you a comparable spec. It has given you a headline.

The capacity-retention threshold problem: 80% vs. 90% SOH (and everything between)

"Cycle life" almost always means "cycles to X% of original capacity," and X is not standardized. Some datasheets test to 80% state of health (SOH), some to 70% or 90%. A cell rated to a lower retention threshold will always show a higher cycle count for the exact same degradation curve, because it's allowed to fade further before the clock stops.

Amprius's own SA112 cell shows this inside one datasheet: 500 cycles across the full 4.2–2.75V voltage window (effectively 100% DoD) to 80% SOH, or 1,000 cycles at 70% DoD to 90% SOH. Same cell, two legitimate numbers, because the test conditions and the retention threshold both changed at once. Read in isolation, "1,000 cycles" and "500 cycles" look like different cells. They're the same cell tested two different ways.

The practical rule: never compare a bare cycle number across two datasheets unless the retention threshold matches. If it isn't stated, assume it isn't 80% and ask.

Test conditions change the number: DoD, C-rate, and temperature

Depth of discharge has the largest effect. Cycling a cell to 100% DoD stresses the electrode structure far more per cycle than cycling it to 50% DoD, so full-DoD cycle counts are always lower than partial-DoD counts on the same cell. Amprius's SA08 cell: 300 cycles at 100% DoD to 80% SOH, versus 700 cycles at 70% DoD to 90% SOH. Two variables moved together there too, but the direction is consistent with every cell in this reference: shallower discharge buys more cycles.

C-rate moves the number even when DoD and the threshold hold constant. Pushing more current through a cell per cycle, on either charge or discharge, accelerates the same mechanical and electrochemical wear mechanisms covered below, so a cycle-life figure measured at a gentle C-rate does not automatically hold at a harder one. A datasheet that states a single C-rate condition is describing one point on that curve, not the cell's behavior across its full current range.

Temperature is the third lever, and it works in both directions. Running warm accelerates the chemical side reactions that consume cycle life (more on that below). Running cold at high current risks a different, faster failure mode entirely. Either way, a cycle-life number measured at 25°C tells you little about performance at -20°C or 60°C, and most datasheets only report the 25°C case.

The Real Degradation Mechanisms Behind Every Cycle-Life Number

Rules of thumb like "deep DoD is worse" are useful for sizing but don't explain why. Three electrochemical mechanisms account for almost all of the capacity and power loss behind a cycle-life curve.

SEI layer growth: the calendar-driven baseline

Every time lithium ions move through the electrolyte, a thin passivation layer, the solid electrolyte interphase (SEI), forms and slowly thickens on the negative electrode. This is happening even when the cell just sits on a shelf, which is why cells also have a separate calendar-life clock running alongside cycle life. SEI growth consumes active lithium and adds impedance over time. It's the background degradation rate every cell experiences regardless of how gently it's used, and it's why two identically-cycled packs can still show different capacity if one spent more total time at high state of charge or high temperature.

Lithium plating: the low-temperature, high-C-rate failure mode

Lithium plating happens when the negative electrode can't intercalate lithium ions fast enough during charging, so metallic lithium deposits on the electrode surface instead of inserting into it. It is most likely at low temperature and high charge current, which is exactly why fast-charging a cold pack is one of the fastest ways to shorten its real-world cycle life regardless of what the datasheet's 25°C number says. Plated lithium doesn't just cost capacity; it can grow dendrites that create an internal short. This is a safety mechanism, not only a lifespan one, and it's the main reason charge current gets derated at low temperature in any pack-level BMS worth using.

Particle cracking and transition-metal dissolution: why deep DoD accelerates wear

Deep discharge and deep charge both push the cathode's crystal lattice through a wider range of expansion and contraction per cycle. Repeated over thousands of cycles, that mechanical stress cracks active-material particles, exposing fresh surface area that reacts with the electrolyte and accelerates SEI growth further. At the same time, operating at high states of charge and elevated temperature promotes dissolution of transition metals (nickel, manganese, cobalt, depending on cathode chemistry) from the cathode, which migrate and interfere with the negative electrode's own SEI layer. Both mechanisms compound with DoD: this is the physical reason the SA08 datasheet shows 700 cycles at 70% DoD against 300 cycles at 100% DoD, and the SA112 datasheet shows 1,000 cycles at 70% DoD against 500 cycles at 100% DoD, on their respective identical cells.

From Duty Cycle to Equivalent Full Cycles: Sizing Pack Lifetime for a Real Mission

Why consumer cycle-life numbers don't apply to drone or robotics duty cycles

A cycle-life datasheet number describes one repeated, uniform pattern: charge to a fixed point, discharge to a fixed point, repeat. A drone flight or a robotics work shift almost never looks like that. A drone might launch at 100% SoC, fly a mission that pulls it down to 55%, land, top up to 90% for a quick turnaround, and repeat with a different depth on the next sortie. None of that maps cleanly onto "one cycle" as the datasheet defines it, so applying the rated cycle count directly to a mission count overstates or understates real lifetime depending on how the real profile compares to the test profile.

A worked example: converting a partial-DoD, high-C pulse-load profile into equivalent full cycles

The standard fix is to convert every partial discharge into an equivalent full cycle (EFC) using Ah throughput: divide the Ah discharged in that event by the cell's rated capacity. A cell that discharges 45% of its rated capacity in one mission has consumed 0.45 equivalent full cycles, regardless of what the pack's state-of-charge display shows before and after.

Take a robotics platform built on the SA112 cell, rated 1,000 cycles at 70% DoD to 90% SOH. The real duty cycle: each work shift discharges the pack from 100% to 55% SoC (a 45% DoD event) under a mixed load with brief pulses toward the cell's 4C pulse limit, then a full CCCV recharge back to 100%.

• EFC consumed per shift: 0.45
• Program requirement: 2 shifts/day, 365 days/year, 2-year service target = 1,460 shifts
• Total EFC over the program: 0.45 × 1,460 = 657 equivalent full cycles
• Rated throughput at the datasheet's own test condition: 1,000 cycles × 70% DoD = 700 equivalent full cycles to 90% SOH

657 EFC against a 700 EFC rated budget leaves headroom, but not much, and the comparison is still an approximation, not a guarantee. Ah-throughput scaling assumes degradation is linear with cycle depth. It isn't: cycle life rises faster than proportionally as DoD drops, so a real 45%-DoD duty cycle will typically outperform this simple ratio, not underperform it. Throughput math is the right tool for a first-pass sizing estimate. Confirming the number for a program that matters means testing at conditions that match the real profile, or building in margin rather than sizing to the theoretical edge.

Pack and BMS-Level Levers That Actually Extend Cycle Life

Every number above describes a cell in isolation. A pack is not a cell; it's a cell plus a BMS plus a mechanical and thermal design, and each of those layers either protects the cell-level cycle-life number or erodes it.

Charge protocol: CCCV taper and current derating by temperature and SoC

A well-specified BMS charges with a constant-current, constant-voltage (CCCV) taper rather than pushing max current all the way to full charge, and it derates charge current at temperature extremes and at high state of charge, exactly where lithium plating risk and cathode stress are highest. This single lever is one of the most effective and least expensive ways to protect the cell-level number, because it's implemented in firmware, not in more expensive cell chemistry. See our custom BMS guide for what a charge-protocol spec should actually include.

Cell balancing and thermal management targets

In a multi-cell pack, the weakest cell sets the pack's real cycle life, not the average cell. Active or passive balancing keeps individual cells inside a tight voltage band so one out-of-spec cell doesn't get driven into deeper DoD or higher stress than the rest of the pack on every cycle. Thermal management matters just as much: keeping the pack inside a controlled temperature window during both charge and discharge avoids stacking the plating risk and the SEI-acceleration risk on top of each other at the same time.

Why pack-level cycle life is always lower than the cell-level datasheet number

Dan-Tech's own Drone Series A pack is rated 1,000 cycles at 80% DoD at the pack level, using Samsung INR21700-50S or Molicel INR21700-P50B cells. That pack-level figure already reflects real balancing tolerance, connector and BMS current handling, and thermal behavior inside an assembled enclosure, none of which shows up in a bare cell datasheet. Treat a cell datasheet's cycle number as a ceiling, and treat a real pack datasheet, when one exists, as the number that actually applies to your program.

Chemistry Selection as a Cycle-Life Decision, Not Trivia

Chemistry sets the ceiling before any pack design decision gets made. The table below states Dan-Tech's own confirmed cylindrical and silicon-anode cell numbers as such. NCA, LFP, and LTO figures are general industry reference ranges, not Dan-Tech-specific claims: Dan-Tech's lineup is built on NMC and silicon-anode (NMC-cathode) cylindrical and pouch cells, and does not currently carry LFP, NCA, or LTO cells.

The trade-off is direct: the chemistries with the highest cycle-life ceilings (LFP, LTO) carry the lowest energy density, and the same pattern holds inside Dan-Tech's own silicon-anode lineup, where cells built for the highest energy density give up some cycle-life ceiling relative to lower-density options in the same family, because energy density and cycle life pull against each other inside a given chemistry family. Choosing a chemistry means choosing which side of that trade-off your program actually needs. For a worked cell-level example of that trade-off, see SA17's 600-cycle rating against its 300 Wh/kg energy density.

How to Read a Supplier's Cycle-Life Claim

Before treating any number as comparable, check for four things. If a supplier's datasheet or sales copy is missing them, the number is not yet a spec.

• No stated depth of discharge. A bare "2,000 cycles" with no DoD is not comparable to anything.
• No stated C-rate. Cycle life measured at 0.5C discharge will not hold at 3C.
• No stated capacity-retention threshold. "Cycles to what?" 70%, 80%, and 90% SOH are three different tests.
• No referenced test standard or temperature. A number with no test standard and no stated temperature is a marketing figure, not an engineering spec.

A bare cycle number with none of that context is not a red flag by itself; every industry has shorthand. It becomes a real problem the moment it's used to compare two different cells or to size a program, without anyone checking whether the conditions behind the two numbers match.

Wichtige Entscheidungen: Zusammenfassung

• Never compare cycle numbers across datasheets without matching the retention threshold, DoD, C-rate, and temperature. A bare number is a headline, not a spec.
• The retention threshold alone (70% vs. 80% vs. 90% SOH) can materially change a stated cycle count for the identical cell and identical degradation curve.
• Degradation comes from three mechanisms: SEI growth, lithium plating, and particle cracking/transition-metal dissolution, and each responds to different levers: calendar time, low-temperature fast charging, and deep DoD, respectively.
• Convert your real duty cycle to equivalent full cycles using Ah throughput before comparing it to any rated cycle count. Treat the result as a first-pass estimate, not a guarantee, since real degradation curves are nonlinear with DoD.
• Charge protocol, balancing, and thermal management are the pack-level levers you actually control. They can preserve or erode the cell-level number by a wide margin.
• A pack-level cycle-life figure, when one exists, is always the more trustworthy number than the bare cell datasheet, because it already reflects balancing tolerance and real assembly conditions.
• Chemistry sets the ceiling. Energy density and cycle life trade against each other inside every chemistry family; pick the side of that trade-off your mission actually needs.

FAQ

What is considered a "good" cycle life for a lithium-ion cell?

It depends entirely on the test conditions behind the number, not the number alone. A cell rated 500 cycles at 100% DoD to 80% SOH under a high C-rate can be a stronger real-world performer than a cell rated 2,000 cycles under a shallow DoD and low C-rate, once both are converted to the same test basis.

Does depth of discharge really matter that much for cycle life?

Yes, more than any other single variable. Cycling to 100% DoD stresses the electrode structure far more per cycle than cycling to 50-70% DoD, and the effect is nonlinear. This is well-established general lithium-ion behavior: many published aging studies show that cutting DoD in roughly half can more than double the achievable cycle count. Dan-Tech's own cell data confirms the same directional trend across the 70-100% DoD range that's actually been tested (for example, SA112 roughly doubles its cycle count going from 100% DoD to 70% DoD).

How many equivalent full cycles does a partial discharge count as?

Divide the Ah discharged in that event by the cell's rated Ah capacity. A discharge from 100% to 55% SoC on a 6.5 Ah cell consumes roughly 0.45 equivalent full cycles, not one full cycle.

Is LFP always the right choice if cycle life is the priority?

Not for every application. LFP's cycle-life ceiling (commonly 3,000-6,000+ cycles to 80% SOH, general industry reference) comes with a real energy-density trade-off. For weight- or volume-constrained platforms like drones, that trade-off often isn't worth it even at a lower cycle count; Dan-Tech does not carry LFP cells for this reason.

Why does a pack-level cycle-life number differ from the cell datasheet?

Balancing tolerance across cells, connector and BMS current handling, and thermal behavior inside an assembled enclosure all affect real pack life and don't show up in a single cell's datasheet. A pack-level figure, when the manufacturer publishes one, already accounts for those factors.

Sizing a Pack Against Your Real Duty Cycle

Cycle life is a program-sizing input, not a spec-sheet trivia point. Getting it right means converting your platform's actual mission profile into equivalent full cycles, choosing a chemistry with the right side of the energy-density/cycle-life trade-off, and specifying the charge protocol, balancing, and thermal management that protect the cell-level number instead of eroding it.

Dan-Tech Energy, custom Li-ion battery pack manufacturer with production in Germany and the US, sizes every pack against the duty cycle it will actually run, not the number on a cell datasheet.

Browse Dan-Tech's Li-Ion pack catalog to see cycle-life-relevant cell options across the current lineup, or use the ToolBox to scope a pack spec against your platform's real duty cycle.