ELECTRONICS + EMBEDDED / POWER PATHS / FIELD GUIDE 011

BATTERY POWER PATHS DESERVE FAILURE DIAGRAMS.

A portable device can have a battery, USB input, charger, system rail, regulator, ideal-diode path, external adapter, and debug cable all capable of moving current. Drawing only the nominal arrows is how reverse current and switchover brownouts become folklore.

CORE IDEAFor every power source and every operating state, draw where current is allowed to flow, where it must not flow, and what happens while the preferred source is changing.

SOURCE OR-ING

Two valid supplies do not become one valid supply by touching them together.

When multiple sources can feed one rail, the design needs a defined source-selection or power-ORing strategy. Simple diode ORing can prevent backfeed but costs forward voltage and power. MOSFET-based ideal-diode controllers can reduce that loss while still enforcing preferred current direction.

The control behavior matters during overlap. If one source is slightly higher, it may carry the load. If the voltages cross slowly, current can migrate between paths. If the controller response is too slow relative to load transients and output capacitance, the system rail can dip during an apparently orderly handoff.

SYSTEM + CHARGER

Charging the battery and powering the load are related but different jobs.

Power-path charger architectures can power the system from an external source while separately managing battery charge current. That avoids forcing all system load through the battery charging path and allows the system to operate even when the battery is deeply discharged or absent, depending on the device architecture.

The charger must allocate available input current among system demand and battery charging. If the load suddenly rises, charge current may need to fall. A design that assumes “USB connected” means “full charge current plus full system load is available” can exceed the adapter, cable, connector, or upstream current limit.

INRUSH

Connecting a supply is a transient event.

Input and system capacitors look like a low impedance when first connected. The resulting inrush can collapse a weak source, trip a USB current limiter, arc a connector, trigger a protection switch, or pull the system through a brownout sequence before the rail ever reaches steady state.

Soft-start, controlled FET turn-on, precharge, current limiting, and properly sized capacitance can shape the event. The correct design depends on the source impedance, load profile, allowable ramp time, and downstream undervoltage behavior.

REVERSE CURRENT

Ask what happens when the output is powered and the input is not.

Regulators, chargers, USB interfaces, GPIO protection structures, and measurement paths can all conduct current backward under conditions the nominal block diagram never shows. An unpowered subsystem can become partially powered through a signal pin. A regulator output held up by another source can feed its input. A USB rail can be unintentionally energized from the battery.

Read reverse-current and absolute-maximum specifications explicitly. “Off” does not mean electrically absent.

SOURCE PRIORITY

Preference needs hysteresis.

If two sources have nearly equal voltage, a naive selector can chatter between them as load changes. Hysteresis, debounce, qualification delay, or controller-specific priority logic prevents repeated switching around the crossover point.

Source policy should also reflect mission. External power may be preferred to preserve battery. A clean regulated source may be preferred over a noisy one. A battery may temporarily support peaks that the adapter cannot. Define the behavior instead of allowing whichever rail wins by a few millivolts to become policy.

SWITCHOVER

The worst rail dip may happen when both sources are healthy.

A system can reset during transfer from battery to adapter or adapter to battery even though both sources measure correctly before and after the event. The cause is the gap between them: FET gate delay, controller qualification, current limiting, source capacitance, connector bounce, or load step.

Capture the system rail at high time resolution while switching sources under worst-case load. A multimeter showing 5.0 V before and 5.0 V after has faithfully missed the event.

USB

A cable is part of the power system.

Cable resistance, connector resistance, negotiated current, source capability, and plug orientation circuitry can shape delivered voltage. A device that works on a bench supply may brown out from a long thin cable at the same nominal input voltage.

Measure voltage at the device input under dynamic load, not only at the source. If USB-C or another negotiated interface is used, treat configuration/negotiation state as part of available-power state.

BATTERY ABSENT + DEAD

Test the states people assume never happen.

Some designs rely on the battery to absorb transient load or stabilize the charger/system node even when external power is present. Removing the battery can reveal startup oscillation, current-limit cycling, or inadequate bulk capacitance.

A deeply discharged battery is a different state from an absent one. Charger preconditioning, pack protection, cell undervoltage cutoff, and system-start requirements can interact. Validate startup with a healthy battery, weak battery, protected-off battery, absent battery, and external power appearing/disappearing in each state.

FAULT CONTAINMENT

One source fault should not automatically poison every rail.

Short circuits, reversed adapters, damaged cables, failed charger FETs, battery protection trips, and external overvoltage all deserve explicit containment. Fuses, e-fuses, current limiting, overvoltage protection, reverse-polarity protection, ideal-diode stages, and domain isolation are architectural tools, not decorative BOM items.

FIELD METHOD

Draw and then force every transition.

1. Make a current-flow diagram for each source state. External only, battery only, both, neither, and fault states.

2. Scope the system rail during hot-plug and unplug. Test at maximum realistic load.

3. Measure reverse current with domains selectively unpowered.

4. Vary cable and source impedance. The adapter label is not the voltage at the board.

5. Test dead and absent battery states separately.

6. Exercise source crossover around priority thresholds. Look for chatter.

7. Correlate resets with rail minima and supervisor thresholds. Preserve the reason the system restarted.

BOTTOM LINE

A power path is a state machine made of copper and FETs.

Every source can appear, disappear, sag, overlap, or fail. Define current direction and recovery for those transitions before the product discovers them in somebody else's hands.

The power diagram should include the arrows you hope never conduct.