5 min read
On this page

Sensor and Actuator Systems

Sensor Types

Temperature Sensors

| Sensor | Type | Range | Accuracy | Interface | |---|---|---|---|---| | DS18B20 | Digital (1-Wire) | -55 to +125C | +/-0.5C | 1-Wire | | LM35 | Analog (voltage) | -55 to +150C | +/-0.5C | Analog | | BME280 | Digital (I2C/SPI) | -40 to +85C | +/-1C | I2C/SPI | | Thermistor (NTC) | Analog (resistance) | -40 to +300C | 1-5% | Analog + divider | | Thermocouple (K) | Analog (voltage) | -200 to +1350C | +/-2C | Analog + amplifier |

Thermistors follow the Steinhart-Hart equation: 1/T = A + B*ln(R) + C*(ln(R))^3

Inertial Measurement Unit (IMU)

An IMU combines multiple motion sensors:

Accelerometer -- measures linear acceleration (including gravity)

  • Principle: MEMS proof mass displacement
  • Output: 3-axis acceleration in g (1g = 9.81 m/s^2)
  • Used for: tilt detection, step counting, vibration monitoring

Gyroscope -- measures angular velocity

  • Principle: Coriolis effect on vibrating MEMS structure
  • Output: 3-axis rotation rate in degrees/second
  • Used for: rotation tracking, stabilization, navigation
  • Suffers from drift over time

Magnetometer -- measures magnetic field

  • Principle: Hall effect or magnetoresistive
  • Output: 3-axis magnetic field in microtesla
  • Used for: compass heading, position sensing
  • Affected by hard/soft iron distortion from nearby metals

Common IMU chips: MPU-6050 (6-axis), MPU-9250 (9-axis), BNO055 (9-axis with onboard fusion), LSM6DSO (6-axis).

Proximity and Distance Sensors

| Sensor | Technology | Range | Resolution | |---|---|---|---| | HC-SR04 | Ultrasonic | 2cm - 4m | ~3mm | | VL53L0X | Time-of-Flight laser | 3cm - 2m | 1mm | | Sharp GP2Y0A21 | Infrared triangulation | 10cm - 80cm | Variable | | LiDAR (TFMini) | Time-of-Flight laser | 0.3 - 12m | 1cm |

Gas Sensors

| Sensor | Detected Gas | Principle | |---|---|---| | MQ-2 | LPG, methane, smoke | Metal oxide semiconductor | | MQ-135 | Air quality (NH3, NOx, CO2) | Metal oxide semiconductor | | SGP30 | TVOC, eCO2 | MOX with I2C digital output | | SCD40 | CO2 (true) | Photoacoustic NDIR | | BME680 | VOC, temp, humidity, pressure | MOX + environmental combo |

MQ-series sensors require a burn-in period and calibration. Digital sensors like SGP30 and SCD40 provide factory-calibrated output.

Reading a Sensor in Embedded Rust

BME280_ADDR ← 0x76

STRUCTURE Bme280Reading
    temperature: real   // Celsius
    pressure: real      // hPa
    humidity: real      // %RH

PROCEDURE READ_BME280(i2c) → Bme280Reading or error
    raw ← array of 8 bytes
    I2C_WRITE_READ(i2c, BME280_ADDR, [0xF7], raw)

    // Apply compensation formulas from datasheet
    // (simplified here -- real code uses calibration data)
    adc_t ← (raw[3] << 12) OR (raw[4] << 4) OR (raw[5] >> 4)
    temperature ← COMPENSATE_TEMPERATURE(adc_t)

    RETURN Bme280Reading(temperature, pressure ← 1013.25, humidity ← 50.0)

Sensor Fusion

Raw sensor data is noisy and incomplete. Sensor fusion combines multiple sensors to produce a more accurate estimate than any single sensor alone.

Complementary Filter

Combines high-frequency data (gyroscope) with low-frequency data (accelerometer) using a simple weighted blend:

angle = alpha * (angle + gyro_rate * dt) + (1 - alpha) * accel_angle

Where alpha is typically 0.95-0.98. The gyroscope provides short-term accuracy (no vibration noise) while the accelerometer prevents long-term drift.

STRUCTURE ComplementaryFilter
    angle: real
    alpha: real

PROCEDURE NEW_FILTER(alpha) → ComplementaryFilter
    RETURN ComplementaryFilter(angle ← 0.0, alpha ← alpha)

PROCEDURE UPDATE(filter, gyro_rate, accel_angle, dt) → real
    filter.angle ← filter.alpha * (filter.angle + gyro_rate * dt)
                  + (1.0 - filter.alpha) * accel_angle
    RETURN filter.angle

Kalman Filter

Optimal recursive estimator that predicts state, then corrects using measurements. Handles noise statistically.

Predict step:

x_pred = A * x + B * u         (state prediction)
P_pred = A * P * A^T + Q       (covariance prediction)

Update step:

K = P_pred * H^T * (H * P_pred * H^T + R)^-1   (Kalman gain)
x = x_pred + K * (z - H * x_pred)                (state update)
P = (I - K * H) * P_pred                         (covariance update)

Where:

  • x = state vector, P = state covariance
  • A = state transition, B = control input, H = observation model
  • Q = process noise, R = measurement noise, K = Kalman gain

The Kalman filter is computationally expensive for embedded (matrix operations). For simple 1D cases, a scalar Kalman filter is practical on an MCU.

Madgwick Filter

An efficient orientation filter specifically designed for IMUs. Computes quaternion orientation using gradient descent optimization.

Advantages over Kalman:

  • No matrix operations -- uses quaternion math
  • Single tuning parameter (beta)
  • Very efficient on Cortex-M (runs in ~300 us at 48 MHz)
  • Handles magnetic distortion gracefully

Used in drones, robotics, motion capture, and VR controllers.

Actuators

Motor Types

| Motor | Control | Characteristics | Applications | |---|---|---|---| | DC Brushed | PWM voltage | Simple, bidirectional, moderate speed | Wheels, fans, pumps | | Stepper | Step/direction pulses | Precise positioning, no feedback needed | 3D printers, CNC | | Servo (RC) | PWM position signal | Built-in feedback, limited rotation | Robotics joints | | BLDC | Electronic commutation | High efficiency, long life, complex control | Drones, EV, hard drives |

H-Bridge Motor Driver

An H-bridge controls DC motor direction using four switches:

    VCC                     VCC
     |                       |
   [S1]---+---Motor---+---[S3]
          |           |
   [S2]---+           +---[S4]
     |                       |
    GND                     GND

Forward:  S1=ON, S4=ON, S2=OFF, S3=OFF
Reverse:  S2=ON, S3=ON, S1=OFF, S4=OFF
Brake:    S1=ON, S2=ON (or S3+S4)
Coast:    All OFF

Common H-bridge ICs: L298N, DRV8833, TB6612FNG. Never turn on S1+S2 or S3+S4 simultaneously (shoot-through destroys the bridge).

Stepper Motor Control

// Simple stepper driver using step/direction interface (e.g., A4988, DRV8825)
STRUCTURE StepperDriver
    step_pin: output pin
    dir_pin: output pin

ASYNC PROCEDURE STEP(driver, direction, steps, delay_us)
    IF direction THEN
        SET_HIGH(driver.dir_pin)
    ELSE
        SET_LOW(driver.dir_pin)

    FOR i ← 1 TO steps DO
        SET_HIGH(driver.step_pin)
        AWAIT DELAY(delay_us microseconds)
        SET_LOW(driver.step_pin)
        AWAIT DELAY(delay_us microseconds)

Servo Motor Control

Standard RC servos accept a 50 Hz PWM signal:

  • 1.0 ms pulse = 0 degrees
  • 1.5 ms pulse = 90 degrees (center)
  • 2.0 ms pulse = 180 degrees
PROCEDURE SET_SERVO_ANGLE(pwm, angle_deg)
    // Map 0-180 degrees to 1.0-2.0 ms pulse at 50 Hz (20 ms period)
    pulse_us ← 1000.0 + (angle_deg / 180.0) * 1000.0
    duty ← (pulse_us / 20000.0) * GET_MAX_DUTY(pwm)
    SET_DUTY(pwm, duty)

Motor Control Algorithms

PID Controller

Proportional-Integral-Derivative control is the workhorse of motor control:

output = Kp * error + Ki * integral(error) + Kd * derivative(error)

STRUCTURE PidController
    kp, ki, kd: real
    integral: real
    prev_error: real
    output_min, output_max: real

PROCEDURE UPDATE(pid, setpoint, measurement, dt) → real
    error ← setpoint - measurement

    // Proportional
    p ← pid.kp * error

    // Integral with anti-windup clamping
    pid.integral ← pid.integral + error * dt
    pid.integral ← CLAMP(pid.integral, pid.output_min / pid.ki,
                                        pid.output_max / pid.ki)
    i ← pid.ki * pid.integral

    // Derivative (on measurement to avoid derivative kick)
    d ← pid.kd * (error - pid.prev_error) / dt
    pid.prev_error ← error

    RETURN CLAMP(p + i + d, pid.output_min, pid.output_max)

Tuning Methods

| Method | Approach | |---|---| | Ziegler-Nichols | Increase Kp until oscillation, derive Ki/Kd from period | | Manual tuning | Set Ki=Kd=0, increase Kp, then add Ki, then Kd | | Cohen-Coon | Based on step response characterization | | Auto-tune | Relay oscillation method (automated) |

Field-Oriented Control (FOC)

Advanced BLDC/PMSM control that transforms 3-phase currents into a rotating reference frame (d-q transform), enabling independent control of torque and flux:

3-phase currents --> Clarke transform --> alpha-beta --> Park transform --> d-q
     (ia, ib, ic)                        (stationary)                    (rotating)

FOC requires:

  • Current sensing (shunt resistors + ADC)
  • Rotor position (encoder or sensorless estimation)
  • Fast control loop (10-20 kHz)
  • Space Vector Modulation (SVM) for inverter switching

FOC delivers smooth, efficient motor operation and is standard in drones, EVs, and industrial drives.

Key Takeaways

  • Sensor selection depends on the measurand, accuracy, interface, and power requirements.
  • Complementary filters are simple and effective for combining gyro and accelerometer data.
  • Kalman filters are optimal but computationally heavy; Madgwick is a practical alternative for orientation.
  • H-bridges enable bidirectional DC motor control; stepper drivers provide open-loop precision.
  • PID control with anti-windup is sufficient for most motor control; FOC is needed for BLDC efficiency.