Live power & battery

The twelve live sensors poll the live_overview endpoint every 30 seconds. All use state_class: measurement, so Long-Term Statistics tracks hourly min/max/mean automatically.

Entity	Key	Unit
PV Power	pv_power	W
Battery Power	battery_power	W (± bidirectional)
Battery SoC	battery_soc	%
Grid Power	grid_power	W (± bidirectional)
Grid Import Power	grid_consumption_power	W (≥ 0)
Grid Export Power	grid_feed_in_power	W (≥ 0)
Total Consumption	consumption_power	W
Household Consumption	household_power	W
EV Charger Power	ev_chargers_power	W
Heat Pump Power	heat_pumps_power	W
AC Power	acs_power	W
Self-Sufficiency	self_sufficiency	%

Self-sufficiency trends

Drop the entity into a statistics-graph card, or build a Statistics Helper for rolling averages. No integration code needed.

Why grid_power is bidirectional

grid_power is the raw signed value from the inverter: positive on import, negative on export. The split grid_consumption_power / grid_feed_in_power channels are the ≥ 0 projections that back the *_energy counters and the Energy Dashboard wiring. Pick the one that matches your downstream consumer.

Battery sensors

• battery_power is bidirectional (positive on discharge, negative on charge in the HA convention used by the Energy Dashboard). Not integrated into kWh directly — see Energy & cost for the split battery_charge_power_energy / battery_discharge_power_energy channels.
• battery_soc is 0–100 %.
• Hybrid-inverter installs (Sungrow, Enphase) carry both entities on the inverter sub-device. FoxESS and other PV-plus-battery-split installs get a separate battery sub-device.