Intelligent Solar Energy Export Controller

Micro Solar AI is a compact, DIN-rail mountable IoT controller built on the ESP32 platform with hardwired Ethernet (LAN8720). It reads real-time power data from inverters and grid meters via Modbus RTU (RS485), integrates solar production and weather forecasts, and intelligently drives up to 4 digital outputs — maximizing self-consumption in residential and commercial PV installations. Unlike simple timer-based or threshold-based controllers, Micro Solar AI uses a predictive approach: it knows how much energy your panels will produce tomorrow and adjusts battery charging, load switching, and export strategies accordingly.

Key Features

Three Independent Modbus RTU Tasks Each task polls a separate Modbus slave device — inverter, grid meter, battery BMS, or any RS485 sensor. Fully configurable: slave ID, register address, function code (FC03/FC04), 16-bit or 32-bit word order, custom scaling (multiply/divide), and export mode for bidirectional meters. Polling intervals from 100ms to 60s per task. A built-in moving average filter (1–64 samples) smooths noisy readings before output decisions. Each output has independent ON/OFF thresholds with hysteresis, minimum ON and OFF timers to protect connected equipment, configurable fail-safe states, and consecutive bad-read counters before declaring a fault. Solar Forecast Integration Connects to the forecast.solar free API and retrieves hourly production estimates for up to 3 independent panel groups — each with its own orientation (azimuth), tilt (declination), and peak power (kWp). The dashboard displays a 48-hour bar chart showing today’s and tomorrow’s expected production. A system-wide efficiency factor (inverter losses, cable losses, dust/shading) adjusts the raw forecast to match real-world output. Automatic retry on API errors with intelligent caching to survive reboots. Weather Forecast Pulls 24-hour weather data from the Open-Meteo API: min/max temperature, hourly wind speed and gusts, cloud cover percentage, humidity, and precipitation. Configurable alerts with thresholds for:

• Frost warning — protect sensitive equipment when temperature drops below threshold
• Heat warning — activate cooling when temperature exceeds limit
• High wind alert — retract awnings or secure outdoor equipment
• Rain/precipitation risk — close greenhouse vents or trigger drainage

Smart Output Control Four digital outputs (via PCF8574 I2C expander) with per-channel configuration:

• Independent ON/OFF trigger thresholds with hysteresis
• Minimum ON time and minimum OFF time (equipment protection)
• Active-high or active-low output polarity
• Fail-safe state (ON or OFF) when communication is lost
• Consecutive fault counter — tolerates brief glitches
• Switch between AUTO (algorithm-driven) and MANUAL mode via web or Modbus TCP

Battery SOC Strategy Reads battery state-of-charge from any Modbus-connected BMS or hybrid inverter. Automatically adjusts the minimum SOC setpoint based on tomorrow’s solar forecast:

• Sunny day expected ? allows deeper discharge (lower min SOC) to maximize solar absorption
• Cloudy day expected ? charges more conservatively (higher min SOC) to ensure backup capacity
• Configurable cloud cover threshold and SOC limits for sunny vs cloudy scenarios
• Optional: writes the calculated min SOC back to the inverter via Modbus register

Data Logging & History 288-entry RAM ring buffer with 5-minute resolution, plus 7-day CSV archive stored on flash (LittleFS). View interactive history charts directly in the browser — no external tools needed. Data survives reboots thanks to automatic filesystem persistence. Log can be cleared remotely via the web interface. Email & SMTP Alerts Three independent alarm channels, each monitoring any input parameter (voltage, temperature, power). Configurable trigger threshold with hysteresis and direction (above/below). State machine tracks alarm lifecycle: NORMAL ? ACTIVE ? CLEARED. Sends formatted email notifications via STARTTLS or SSL with bilingual templates (English and Romanian). MQTT Publishing Publishes complete device status as a JSON payload to any MQTT broker (Mosquitto, HiveMQ, AWS IoT, etc.) at configurable intervals. Supports authenticated connections with username/password. Perfect for integration with Home Assistant, Node-RED, Grafana, or custom dashboards. Zabbix Monitoring Native Zabbix sender protocol support — pushes key-value pairs directly to a Zabbix server or proxy. Ideal for enterprise environments where centralized monitoring and alerting infrastructure is already in place. Configuration Import/Export Full device configuration can be exported as a JSON file (backup) and imported on the same or different unit (restore/clone). Makes it easy to deploy multiple units with identical settings or recover from a factory reset.

Connectivity

Web Interface

Dashboard (Port 80) Real-time overview of all Modbus readings (raw, scaled, averaged, export watts), output states, failure counters, and error messages. Solar forecast with 48-hour bar chart. Weather forecast with temperature, wind, cloud, and precipitation data. History tab with interactive 7-day charts. Info tab with device details, network status, and uptime. All data updates live via WebSocket — no page refresh needed. Configuration & Control (Port 8080) Tabbed interface with two main views:

• Configuration tab — all device settings: network, Modbus tasks, thresholds, solar panels, weather alerts, MQTT, Zabbix, email, alarms, authentication. Save, Reboot, Export, and Import buttons.
• Control tab — live output control: AUTO/MANUAL mode switch, individual output toggles, ALL ON/ALL OFF, live status grid (firmware, uptime, network, IP, free heap, temperature, task values, solar production), system actions (Reboot, Clear Log, Solar Reset).

Bilingual user interface: English and Romanian, switchable from the configuration page.

Security

• Port separation — monitoring (port 80) and control (port 8080) use independent credentials
• HTTP Basic Auth on all ports — no anonymous access
• OTA protected — firmware upload requires control credentials
• Read-only dashboard — port 80 cannot modify outputs, reboot, or clear logs

Hardware Specifications

Use Cases

• Solar self-consumption optimization — read the grid meter and automatically switch loads (water heater, pool pump, EV charger) to absorb excess solar production instead of exporting to the grid
• Zero-export control — keep grid export at zero by managing multiple loads in priority order based on available surplus power
• Battery strategy optimization — adjust battery min SOC based on tomorrow’s weather forecast, maximizing solar utilization while maintaining backup capacity
• Greenhouse automation — combine temperature sensor + weather forecast for heating/cooling and frost protection
• Industrial monitoring — read any Modbus RTU device and publish data to MQTT/Zabbix with configurable alerts
• Multi-inverter installations — monitor up to 3 Modbus devices simultaneously with independent polling and scaling

What’s in the Box

• 1x Micro Solar AI controller (DIN-rail mount)
• 1x DS18B20 temperature sensor with cable
• 1x Quick start guide