Arduino-Based Water Quality Monitoring with pH, Turbidity, Temperature, LDR, and Water Level Sensors

Digital filtering eliminates noise from sensor signals or smooths trends. Moving average calculates input averages over time intervals: fixed averages compute results after collecting N points, while dynamic averages recalculate at each sample by including new and excluding oldest points. A key trade-off exists: more points improve smoothing but increase delay. Memory requirements grow with window size, potentially causing resource issues.

The average filter demonstration shows raw noisy data (red) and filtered data (blue) converging to the true value (14.4). This demonstrates how recursive average filtering removes noise while tracking the underlying signal. Moving average filter extends simple average by maintaining a fixed window size. Unlike simple average which considers all data, moving average only considers the most recent N data points. When new data arrives, the oldest data point is discarded to maintain constant window size. The formula is: X_k = (1/N) * Σ_{i=0}^{N-1} X_{k-i}. This allows tracking changing signals while reducing noise, making it suitable for applications where the underlying signal changes over time, such as stock market analysis.

This section establishes the foundational motivation for Kalman filtering in autonomous systems. Raw sensors suffer from noise, bias, and drift—GPS errors range from 1-10m depending on conditions, while IMUs accumulate bias over time. Single sensors cannot observe all state variables, and nonlinear sensors add complexity. The Kalman filter provides a statistically optimal recursive solution to estimate hidden states by combining multiple noisy sensor measurements with system models. For linear systems, the mathematical framework consists of the process equation x_dot = Ax + Bu and measurement equation z = Hx, discretized as x_k = Ax_{k-1} + Bu_{k-1} + w_k and z_k = Hx_k + v_k, where w_k and v_k represent zero-mean Gaussian process and measurement noise with covariances Q and R respectively.

Navigation sensors accumulate small errors over time. A gyroscope measuring 0.01 degrees per second when the drone is actually flying straight adds up to 36 degrees of phantom rotation over an hour. An accelerometer with a tiny bias of 0.001g accumulates as phantom velocity and then phantom distance. The navigation computer uses a Kalman filter to minimize these errors by continuously weighing each sensor's reliability against others and calculating the most probable position. However, even the best filters cannot eliminate drift entirely—position uncertainty can grow to 5-8 km over 90 minutes of autonomous flight.

This tutorial demonstrates how to combine data from MPU6050 accelerometer-gyroscope and HMC5883L magnetometer sensors using sensor fusion and Kalman filter algorithms to accurately measure 3D orientation angles (pitch, roll, and yaw) on an ESP32 microcontroller, overcoming individual sensor limitations such as noise, drift, and magnetic interference through mathematical filtering techniques.

Microcontrollers like ESP32 feature built-in DeepSleep modes that dramatically reduce power consumption from milliamperes to microamperes. For maximum efficiency, the Power LED should be disconnected during sleep. For even greater power savings, external RTC (Real-Time Clock) circuits can wake the system at predetermined intervals, allowing the entire microcontroller and voltage regulation circuitry to remain in sleep mode until needed. This approach is particularly valuable when minimizing battery size or maximizing deployment duration is critical.

Solar-powered devices implement sophisticated power management strategies. The microcontroller controls power distribution to sensors, LEDs, and the battery. Lithium batteries cannot be charged below freezing temperatures, as lithium ions cannot intercalate properly and form damaging layers. The solar panel generates 5.5 volts through 11 sections, fed through a 100 ohm resistor to an NPN transistor for voltage regulation. The microcontroller monitors voltage through a 100K resistor divider to detect dusk and manage charging. When the cell reaches 4.2 volts, the microcontroller shunts the voltage down and turns off charging. This demonstrates how embedded systems implement environmental sensing and power management for efficient operation.

Microcontroller power management enables years-long battery operation through multiple sleep modes: light sleep, deep sleep, and hibernation. Deep sleep can reduce power to microampere levels, achieving 7-10 year battery life in properly designed systems. Edge computing principles recommend preliminary data processing at the sensor level to minimize communication overhead. Power transitions between sleep modes require timing optimization—applications waking from deep sleep must account for transition delays when scheduling measurements.

This extensive demonstration covers the complete process of designing an embedded monitoring system for battery applications. The project uses an STM32 L073 microcontroller (Cortex-M0+ at 32MHz, 192KB flash, 20KB SRAM) as the central processing unit, demonstrating how to select appropriate microcontrollers based on application requirements. Key topics include: hardware integration of multiple sensors (INA219 for current/voltage, TMP36 for temperature, and Sharp memory LCD for display), power optimization techniques including voltage regulation and sleep modes, and system-level power analysis comparing custom solutions against commercial alternatives. The project demonstrates achieving approximately 0.34W active power consumption and near-zero sleep mode consumption, representing approximately 4x improvement over commercial modules.

A properly configured microcontroller in sleep mode can consume as little as 25 microamps. This extremely low power consumption is achieved by disabling most peripheral functions and allowing the watchdog timer to handle wake-up. At this power level, a typical battery can power the device for many weeks, making it suitable for long-term battery-powered applications.

In Oxygen Not Included, temperature sensors can trigger automated valve systems. The player sets a 50°C threshold: when temperature exceeds 50°C, the valve closes and the cooler activates; when below 50°C, the valve opens and the cooler deactivates. This creates an automated temperature regulation loop. The player also explains liquid priority systems (pontes de líquido) that determine which water flows first through junctions, ensuring turbine water has priority to flow through the cooling system.

Three sensors manage water: flow sensor measures water flow, death sensor measures water death, and contamination sensor measures contamination levels. The valve (formerly snooze) can be combined with contamination sensors to automatically close when contamination exceeds thresholds like 5%. The death sensor can automate fluid dumps based on water levels, with players setting thresholds (like 0.4 meters) to trigger actions when levels fall below. This enables automated irrigation and water management systems.

Sensors are devices that detect physical changes in the environment (such as light, temperature, or sound) and send this information to a processor like Arduino, which then makes decisions and triggers actuators—devices that convert electrical energy into physical actions (such as motors, pumps, or valves)—to automate processes and create interactive systems.

Automation systems can be configured to trigger actions when values fall below or exceed specific thresholds. For example, a pump can be activated when water depth is less than 1 unit, and irrigation can be controlled when depth is less than 5 units. This threshold-based approach allows for precise control over automated systems based on real-time measurements.

Automated water quality control systems use sensor-based feedback loops to maintain optimal aquatic conditions. The system implements tiered responses based on turbidity readings: clear water (0-20 NTU) requires no pump activation, moderately cloudy water (21-50 NTU) triggers 5-minute pump cycles, and very cloudy water (>50 NTU) activates 10-minute pump cycles. Similarly, pH control uses threshold-based logic where pumps activate when readings fall below or exceed optimal ranges, creating closed-loop control systems that automatically respond to environmental changes without human intervention.