Controller Battery

The controller battery keeps the SRAM inside the cabinet powered when the main disconnect is off. That SRAM holds TP programs, position registers, KAREL variables, the DCS configuration on capable systems, and the mastering data the axes need to know where they are at power-up.

It's a separate part from the encoder battery on the arm. The controller battery protects controller memory; the encoder battery protects the pulsecoder position counter. Losing one does not lose the other, and losing this one means restoring from the last SRAM backup and re-entering any site-specific data changed since.

Manufacturers publish a service life in years, typically two to four depending on model, and recommend replacing on a schedule tied to a preventive-maintenance cycle. Waiting for the SYST battery-low warning is risky business: if the battery dies completely before you get to it, SRAM contents are gone and the site needs its last known-good backup.

Selection is straightforward. Order the FANUC-specified lithium pack (usually a 3.0 V or 6.0 V unit depending on cabinet), and steer clear of a generic AA holder with alkaline cells. Voltage and self-discharge behavior are matched to the SRAM leakage current, and the wrong cell chemistry will drop the memory even if the label voltage looks correct.

Compatibility limits: the battery lead is keyed and locked to the controller PSU or the CPU board depending on cabinet generation. Some cabinets route the cell through an external battery-backup unit mounted on the door instead of clipping directly to the board. Confirm the connector style before pulling the old pack.

Replacement is done live. Power the cabinet on, back up the SRAM to a USB stick, disconnect the old battery, snap the new one in within the connector's specified hold-up window (documented per model, typically a bit under an hour on standard controller capacitance), and confirm no low-battery warning at the next power cycle. Skip the backup step at your peril.

Practical example: a shop that ran a robot for six years with the same battery cold-started one morning to a bare controller (no programs, no registers, no mastering) because the pack had died a year earlier and nobody logged the warning. Restoration from an old image took three hours of setup plus an hour per axis for zero-position mastering with the single-axis mastering guide.