Cold-start behaviour at −40 °C

−40 °C is a specification, not a demonstration

A cold-start rating is only meaningful when it comes with an envelope attached. Our prismatic LIC series are specified from −40 °C to +70 °C depending on the series; pouch-format cells sit in a narrower −25 … +55 °C band, which is why the format has to be chosen against the site rather than against the datasheet headline. Storage is rated −40 … +80 °C with 0 … +35 °C recommended for long dormancy.

That envelope is what makes an unheated enclosure a design decision rather than a risk. Wind pitch cabinets, outdoor telecom and metering enclosures, rail trackside equipment, port machinery and cold-storage halls all share the same problem: the backup has to work on the coldest morning of the year, after weeks of doing nothing, with no operator present and no power available to warm anything up.

Why LIC chemistry behaves differently in the cold

The difference is mechanical before it is chemical. A lithium-ion battery stores charge by inserting lithium ions into a host structure, and both the insertion and the extraction slow down sharply as the electrolyte thickens and the surface kinetics stall. Charging below roughly 0 °C risks depositing lithium metal instead of intercalating it, which is why battery systems restrict charge current in the cold; usable capacity then collapses toward −20 °C, and lead-acid does the same thing for its own reasons as electrolyte viscosity rises.

An LIC stores charge by physical adsorption at a high-surface-area carbon electrode, with a pre-lithiated negative electrode that lifts the cell voltage to 3.8/4.0 V. Charge sits at the surface rather than inside a host lattice, so the cold does not have to unlock a diffusion path before the cell can deliver. What the cold does affect is the electrolyte's ionic conductivity, and that shows up as resistance rather than as a loss of stored charge. The practical consequence is the sentence that matters to a specifier: at −40 °C the module has more resistance but not much less charge, so it delivers cranking-class current — just with more sag and more heat than it would at 25 °C.

The second cold-weather property is safety rather than performance. Charge stored by physical adsorption has no thermal-runaway mode in the chemistry, so a cold module that is asked for a hard current does not carry the abuse risk that a cold battery pack does when it is pushed outside its charge window. Ninety metres up a tower, that is worth more than a datasheet decimal.

ESR rise with temperature, and how to plan for it

Both capacitance and ESR move with temperature, and they move in opposite directions. Capacitance falls modestly — a planning figure of roughly 10–20% down at −40 °C against 25 °C — while ESR rises by a much larger multiple, because the resistance is dominated by the electrolyte. The bands below are what we use to open a cold-site sizing calculation before the cell series' own characterization curve is applied; they are planning values, and the curve for your chosen series is the number that goes in the design file.

The reason ESR deserves its own row is that a cold load is usually current-limited rather than energy-limited. A pitch motor that needs 300 A to start a feathering sequence will take that current at 25 °C without complaint; at −40 °C the same motor demand meets a string whose resistance has grown several times over, and the resulting sag — ΔV = I × ESR — is what decides whether the drive starts or trips. Energy that is still sitting in the capacitors is no help if the voltage at the terminals has fallen below the load's minimum.

Cold sizing, worked through

Take a 24 V-class string of six 3.8 V cells: 22.8 V at full charge, an 18 V floor, and a feathering load that needs about 20 kJ with a 300 A motor inrush. Two checks have to pass, and at −40 °C they usually disagree about how much hardware is needed.

Read the last three rows as the real lesson of cold-site design. The energy calculation asks for about 510 F of 22.8 V-class string capacitance; the cold-sag calculation says one such string only just holds the floor at the coldest start, with no margin for an aged string or a worse-than-planned inrush. Two strings in parallel fix the sag and also halve the per-string current, which lowers the heat and slows the aging that made the margin thin in the first place. In cold designs, the parallel count is usually set by the sag check and the energy arrives as a by-product.

Standby, dormancy and the heater you no longer need

Cold sites are hard on backup because the equipment is idle for long stretches between events. A pitch cabinet may sit for weeks; a trackside cabinet may sit for months between failures. Two published figures carry that duty: a cell holds ≥3.7 V after a 72-hour open-circuit rest, and the module's management board draws under 20 µA in sleep. Modules with µA-class dormancy measured on the finished unit are what make a site visitable once a year instead of once a season.

This is also where a battery installation accumulates parts. Cold battery banks get heater mats, thermostats, insulation and the control logic that decides when to warm the pack — plus the parasitic load and the extra failure modes those parts bring, all in an enclosure that is already hard to service. LIC needs none of it, and needs no low-temperature charge restriction either, because there is no intercalation step to protect. Removing the heater removes a maintenance item, a standby drain and a reason for the backup to be offline.

Specifying for an outdoor cabinet or an unheated hall

Cold specifications fail when they name a temperature without naming the conditions around it. Nine lines close the gap.

• Coldest start temperature the site can present, not the average winter ambient.
• Minimum bus voltage at which the load still operates, which is the number the sag check is measured against.
• Peak current and duration at that temperature — inrush, not steady state.
• Energy per event and how many events may occur back to back on a cold day.
• Duty cycle, since a shallow sweep protects the cycle-life class better than a deep one.
• Dormancy pattern: days or weeks between events, and how long the site is unattended.
• Airflow and mounting: forced-air builds keep 3 mm ducts between cells, and a cabinet full of insulation changes the thermal picture.
• Insulation requirement, proven at AC 2000 V for one minute on the finished module.
• Telemetry expectations, so a cold sag can be read as cold resistance rather than as a failing cell.

The evidence that backs a cold claim

Any supplier can print −40 °C on a datasheet. Three documents make it real. First, a temperature characterization run that sweeps capacitance and ESR across the −40 … +70 °C envelope in a chamber, so the derating curve is measured rather than asserted. Second, a cold-start current test at the coldest applicable ambient, with the resulting sag recorded — this is what proves the cranking claim on the actual assembly rather than on a single cell. Third, the per-module test report that ships in the carton, which gives you the beginning-of-life capacitance and DC ESR of the unit in your cabinet to compare against once it has been through a winter.

Put those three next to the temperature channels publishing over CAN and the cold-weather question becomes an engineering record with numbers attached rather than a claim you have to take on faith. If the site is cold and the bus is non-standard, the 16–108 V ODM route builds the string to your voltage on the same sorted, laser-welded line — and the wind pitch note walks the same argument through a nacelle installation.