📊 Device Information

Internal chip temperature. If you see values above 100°C, your house is probably on fire. RUN!

Total restarts since power-up. A reset is consecutive if it happens in less than 180 seconds after the previous one. A high number of consecutive resets indicate a potential system instability.

Firmware panics or exceptions that caused an uncontrolled restart. Even one crash is worth investigating via the log page.

Do not worry if this gets too low - the firmware will automatically handle restarts.

Flash filesystem storing energy data, and log files. Logs are limited to a size of around 200 kB, but in the long run the energy data may fill up the storage. You can interact with this via API.

The ADE7953 energy meter chip signals the ESP32S3 via an interrupt pin approximately every 200 ms when new measurement data is ready.

Line-cycle (CYCEND) interrupts serviced for meter readings. Lower than Total Interrupts, which also counts the per-cycle zero-crossing (ZX) interrupts used for grid frequency.

Voltage zero-crossing interrupts, one per line cycle, driving the grid frequency measurement.

Interrupt status snapshots serviced. Total Interrupts minus this is how many ISR wakeups were coalesced into another pass (expected under load; the status snapshot covers every pending cause either way).

Interrupt status snapshots with none of the known bits (ZX, line-cycle end, reset, CRC change) set. Should stay at zero; a non-zero count means the chip raised a cause the firmware doesn't recognize.

Valid energy measurements accepted and processed. A reading is considered successful when SPI communication is verified and the measured values are within expected ranges. For each 2 interrupts we have 3 readings (two for channel 0, one for the other channels).

A failed reading occurs when the measurements are invalid (e.g. negative power for loads or PV) or when the SPI communication verification identifies a mismatch. A very small percentage of failed readings is expected due to the thorough verification process.

Total custom MQTT messages published since boot. Errors indicate messages that could not be delivered to the broker (connection lost, broker rejected, etc.). Check the logs for details on MQTT publish errors.

Meter readings evicted from the cloud publish queue because it was full. The oldest queued point is dropped to make room for the newest. A non-zero count means the device is producing readings faster than they can be published.

Grid frequency/voltage points evicted from the cloud publish queue because it was full. The oldest queued point is dropped to make room for the newest.

HTTP requests received by the ESP32S3 from other devices on the network. Errors indicate unauthorized or malformed requests, or unsupported endpoints. Check the logs for details on request errors.

Measurement batches sent to InfluxDB. Errors indicate upload failures (network issue, wrong token, bucket not found). The device retries on the next interval. Check the logs for more details.

Modbus TCP requests received from other devices on the network. Errors indicate malformed requests or unsupported function codes. Check the logs for details on request errors.

Log entries lost because the internal log buffer was full. A non-zero count means the device could not keep up with the logging rate.