2026-07-16
When your main power drops, downstream logic and memory don't care whose fault it is — they just brownout and corrupt. A battery backup switchover circuit keeps the rail alive by handing off to a battery (or supercap) fast enough that the load never notices. The trick is doing it without glitches, reverse-charging the battery, or leaking main-rail current into the cell when everything is fine.
The naive approach — two Schottky diodes OR-ing main and battery — works but wastes 300-400 mV and drains the battery whenever main sags below (V_bat − V_f). For a 3 V coin cell backing up a 3.3 V rail, you're already discharging the cell during normal operation. Bad.
Better: a dedicated backup supervisor IC (TI TPS3701, Maxim MAX16054, Analog Devices ADM8611). These monitor V_main against a threshold and actively switch a P-channel MOSFET pair. The main-side FET conducts normally; when V_main falls below (V_bat + hysteresis), the controller flips within microseconds to the battery-side FET. RDS(on) losses replace diode drops — typically <50 mV at 10 mA loads.
Key design constraints:
Real-world example: RTC backup on an industrial data logger. The DS3231 RTC draws ~1.5 µA in timekeeping mode. A CR2032 (220 mAh) backing it through an ideal-diode switchover lasts ~16 years theoretically. Through a Schottky with 100 nA reverse leakage and 300 mV forward drop during power-loss windows, that shrinks to 5-6 years. The active switchover pays for itself in field lifetime.
Rule of thumb: Size your hold-up cap so C ≥ (I_load × t_switchover) / (V_nominal − V_brownout). For 50 mA load, 20 µs switchover, and 300 mV droop budget: C ≥ 3.3 µF. Add 3-5× margin for ESR and temperature — call it 22 µF.
Don't forget to add a small (10-100 Ω) resistor in series with the battery to limit inrush if the backup cap is deeply discharged at startup.
