Sensors, Transducers, Actuators, and Electrical Measurement

Introduction

IoT and smart electrical systems depend on a measurement chain that converts physical conditions into trustworthy digital information and, when necessary, converts decisions back into physical action. A voltage sensor measures an electrical potential, a current sensor measures charge flow, a temperature sensor measures thermal state, and an actuator changes the operating condition of equipment. Between these functions lie signal conditioning, isolation, analog-to-digital conversion, communication, calibration, and data interpretation.

A sensor is a device that responds to a physical quantity, while a transducer converts energy or information from one form to another. In practice, the terms are often used interchangeably, although a transducer may include both the sensing element and its conversion circuitry. An actuator is an output transducer: it receives an electrical or digital command and produces motion, switching, heat, light, or another physical effect.

The quality of an IoT system is limited by its weakest measurement link. A highly sophisticated cloud algorithm cannot correct an incorrectly selected current transformer, inadequate isolation, aliasing, sensor drift, or an uncalibrated energy meter. Consequently, sensor selection is an engineering activity involving accuracy, safety, bandwidth, loading, environmental conditions, maintainability, cost, and cybersecurity.

Learning Objectives

After completing this chapter, the reader should be able to:

1. Distinguish sensors, transducers, actuators, measurement systems, and data-acquisition systems.
2. Select suitable sensors for voltage, current, power, energy, temperature, vibration, light, occupancy, and environmental variables.
3. Calculate basic measurement quantities and quantify error, uncertainty, resolution, and sampling requirements.
4. Design a suitable signal-conditioning and isolation path for an IoT node.
5. Explain calibration, traceability, metrology, and uncertainty budgets.
6. Interface sensors and actuators safely with microcontrollers and communication networks.
7. Analyze an applied smart-electrical-system measurement problem.

1. The Measurement Chain

A measurement system can be represented as:

measurand → sensing element → transduction → signal conditioning → sampling and conversion → communication/storage → decision or control → actuator

The measurand is the physical quantity intended to be measured. The sensing element interacts with it. Signal conditioning may amplify, attenuate, filter, linearize, rectify, or electrically isolate the signal. An analog-to-digital converter (ADC) samples the conditioned signal and expresses it as a numerical code. Software then estimates a useful quantity, such as root-mean-square voltage, energy consumption, or occupancy state.

A static sensor characteristic is often approximated by:

where x is the measurand, y is the output, a is sensitivity, and b is offset. This linear model is valid only over the specified operating range. Real sensors may exhibit nonlinearity, hysteresis, saturation, temperature dependence, noise, and ageing.

A measurement result should be reported as an estimate and its uncertainty, not as an unquestionable value. For example, 230.4 V ± 0.8 V communicates more information than 230.4 V alone.

Important measurement characteristics

Accuracy and precision are not synonyms. A sensor may be precise but inaccurate if it repeatedly gives a value displaced from the reference. Calibration can correct systematic offset and gain errors, but it cannot remove random noise completely.

2. Sensors and Transducers for Electrical Quantities

2.1 Voltage measurement

A voltage sensor must present sufficiently high input impedance so that it does not significantly load the circuit. For a resistive divider,

where Vᵢ is the input voltage, Vₒ is the divider output, and R₁ and R₂ are series resistors, with R₂ connected from the output node to the reference point. This equation assumes ideal resistors and negligible loading by the following circuit.

The divider ratio is selected so that the maximum expected input, including transients, remains below the ADC input limit. Resistor voltage ratings, power dissipation, tolerance, creepage, clearance, and surge protection must also be considered. For mains systems, a simple divider connected directly to a microcontroller is unsafe unless the entire low-voltage circuit is appropriately isolated and designed for the relevant safety category.

Common alternatives include isolation amplifiers, isolated voltage transducers, potential transformers, and certified energy-metering integrated circuits. A potential transformer provides galvanic isolation but has ratio, phase, frequency, and burden limitations.

2.2 Current measurement

A shunt resistor produces a voltage proportional to current:

where Vₛ is shunt voltage, I is current, and Rₛ is shunt resistance. Its power dissipation is:

The shunt offers good linearity and bandwidth, but it introduces power loss and may place the sensing circuit at a hazardous common-mode voltage. Kelvin connections reduce errors caused by lead resistance.

A current transformer (CT) provides isolation for alternating current. Its secondary current is approximately proportional to primary current, subject to turns ratio, core behaviour, burden resistance, frequency, and phase error. A CT secondary must not be left open-circuit while current flows in the primary; a potentially dangerous voltage can develop. A burden resistor converts secondary current to a measurable voltage.

Hall-effect sensors measure magnetic field and can measure AC and DC current while providing isolation. Open-loop Hall sensors are convenient but generally less accurate than closed-loop or fluxgate designs. Rogowski coils produce a voltage proportional to the time derivative of current and therefore require an integrator; they are useful for large, rapidly changing AC currents but do not directly measure DC.

2.3 Power and energy measurement

For instantaneous single-phase real power,

where v(t) and i(t) are instantaneous voltage and current. The average real power over an interval T is:

For sinusoidal voltage and current, the familiar expression is:

where φ is the phase angle between the fundamental voltage and current. This sinusoidal expression should not be used blindly with distorted waveforms; sampled instantaneous multiplication is more generally applicable.

Electrical energy is the integral of power:

If power is constant, E = Pt. Energy meters commonly report watt-hours or kilowatt-hours. One kilowatt-hour equals 3.6 megajoules.

In a balanced three-phase sinusoidal system, total real power is:

where Vᴸ and Iᴸ are line-to-line voltage and line current. The expression assumes a balanced system and sinusoidal quantities. In unbalanced or distorted systems, phase-by-phase sampling and summation is safer.

Power-factor measurement must distinguish displacement power factor, associated with phase angle, from true power factor, which also includes waveform distortion. An IoT meter should specify sampling rate, analog bandwidth, phase matching between voltage and current channels, isolation, and accuracy under non-sinusoidal loads.

Worked example: single-phase energy

A load operates at 230 V rms and 8.0 A rms with a power factor of 0.85 for 5 hours. Assuming sinusoidal steady-state operation,

The energy is:

The result excludes standby changes and measurement uncertainty. A real meter would integrate sampled power over time rather than assume constant values.

3. Sensors for Physical and Environmental Variables

3.1 Temperature

Resistance temperature detectors (RTDs), thermistors, thermocouples, and integrated semiconductor sensors are widely used. A platinum RTD such as Pt100 offers stability and approximate linearity; a thermistor provides high sensitivity but strongly nonlinear resistance; a thermocouple covers broad temperature ranges and is rugged but produces a small voltage and requires cold-junction compensation.

For a simple voltage divider using a resistive sensor,

where Rᵀ is sensor resistance, R_fixed is the fixed resistor, and Vₛ is the excitation voltage. Excitation must be controlled because self-heating can bias the result. Temperature sensors should be placed where the desired temperature exists, not merely where installation is convenient.

3.2 Vibration

Piezoelectric accelerometers are suitable for dynamic vibration because they have broad bandwidth, but they are poor choices for static acceleration. MEMS capacitive accelerometers can measure static tilt and low-frequency motion and are readily interfaced digitally. Sensor mounting, axis alignment, resonance, cable motion, and sampling frequency strongly affect the result.

Acceleration data may be summarized using root-mean-square value:

where aₙ is the nth sampled acceleration and N is the number of samples. This estimate assumes a suitably selected sampling window and removal of unwanted bias. Vibration diagnosis may also use frequency spectra, crest factor, peak value, and envelope analysis.

3.3 Light

Photodiodes provide fast, approximately proportional photocurrent over a specified range. Phototransistors provide greater sensitivity but are slower and less linear. Light-dependent resistors are inexpensive but have slow response, broad spectral sensitivity, and significant variation. A lux sensor with a calibrated photometric response is preferable when illumination must be compared with lighting standards.

A transimpedance amplifier converts photodiode current to voltage:

where Iₚₕ is photocurrent and R_f is feedback resistance. Stability requires attention to photodiode capacitance and amplifier bandwidth.

3.4 Occupancy and motion

Passive infrared (PIR) sensors detect changes in infrared radiation and are effective for movement detection, but they may fail to identify a stationary person. Microwave and millimetre-wave radar sensors can detect motion and, in some systems, presence with greater sensitivity. Ultrasonic sensors estimate distance from acoustic time of flight. Cameras can provide rich occupancy information but introduce privacy, bandwidth, lighting, and cybersecurity concerns.

Occupancy systems should define whether the requirement is motion, presence, people counting, or location. A sensor selected for motion detection cannot automatically satisfy continuous-presence detection.

3.5 Environmental variables

Humidity may be measured using capacitive or resistive polymer sensors. Pressure sensors commonly use piezoresistive or MEMS structures. Air quality systems may use metal-oxide, electrochemical, nondispersive infrared, or photoionization methods, each with different selectivity, warm-up time, cross-sensitivity, and maintenance requirements. Carbon-dioxide measurement for ventilation control is often based on nondispersive infrared sensing, but installation must avoid direct exhalation and poor air circulation.

Environmental sensors are particularly sensitive to placement. A temperature sensor near a power converter measures the converter’s thermal field rather than room air. A humidity sensor exposed to condensation or chemical vapour may show accelerated drift. Datasheet accuracy is meaningful only under the stated temperature, humidity, supply, airflow, and ageing conditions.

4. Signal Conditioning, Isolation, and Data Acquisition

4.1 Signal conditioning

Typical conditioning functions include:

• amplification of low-level signals;
• attenuation of signals exceeding ADC range;
• buffering to prevent loading;
• filtering of unwanted frequency components;
• rectification or demodulation where appropriate;
• linearization and cold-junction compensation;
• protection against electrostatic discharge and surges; and
• level translation between sensor and processor voltage domains.

An anti-aliasing low-pass filter is required before sampling when signal energy exists above half the sampling frequency. The Nyquist condition is:

where fₛ is sampling frequency and fₘₐₓ is the highest significant input frequency. In practice, sampling must be substantially higher than twice the nominal frequency because a filter needs a transition band. Aliasing cannot generally be removed after sampling.

An ideal N-bit ADC with input span V_range has an approximate quantization step:

The ideal quantization error is approximately ±q/2, assuming rounding and a uniformly distributed signal relative to code boundaries. Real effective resolution is lower because of noise, nonlinearity, reference error, and clock or aperture effects.

4.2 Isolation and safety

Isolation may be achieved using transformers, optocouplers, capacitive or magnetic digital isolators, isolated amplifiers, or isolated power supplies. Isolation is not merely a signal property: creepage, clearance, insulation rating, surge withstand, grounding, enclosure, and connector design must all be evaluated.

For mains-connected systems, designers must apply the relevant national wiring rules, equipment safety standards, electromagnetic-compatibility requirements, and measurement-category ratings. A low-voltage label on a circuit board does not make its input safe. Protective earthing, fusing, current limitation, touch protection, and safe test procedures are essential.

4.3 Sampling and synchronization

Sampling interval determines temporal resolution, but measurement quality also depends on synchronization. Voltage and current channels used for power measurement must have matched timing and phase response. A phase error produces power error, especially at low power factor.

Sampling clocks should be stable enough for the application. For energy measurement, timestamp accuracy, missing samples, reset behaviour, and data gaps must be addressed. Edge devices should retain raw or diagnostic data when communications fail, while avoiding unnecessary storage of sensitive information.

5. Calibration, Metrology, and Uncertainty

Calibration compares an instrument with a reference under defined conditions. It does not necessarily mean adjustment. A calibration certificate should identify the instrument, method, reference standards, environmental conditions, results, uncertainty, and traceability chain.

A two-point linear calibration can be represented as:

where a is measured gain and b is measured offset in the model y_raw = ax + b. Multi-point calibration or a lookup table may be needed for nonlinear sensors. Calibration must cover the operating range and include the complete signal path where practical.

The combined standard uncertainty for independent input quantities is commonly estimated by the root-sum-square relationship:

This form assumes the inputs are uncorrelated and expressed as standard uncertainties. If quantities are correlated, covariance terms are required. An expanded uncertainty is often reported as U = k u_c, where k is a coverage factor selected for the desired coverage probability and statistical assumptions.

Worked uncertainty example

Suppose a voltage measurement has independent standard uncertainties of 0.40 V from the sensor, 0.25 V from the ADC/reference, and 0.15 V from calibration repeatability. Then:

For an illustrative coverage factor k = 2, the expanded uncertainty is approximately ±0.99 V. The independence assumption must be checked; shared temperature effects, for example, could create correlation.

Metrology requires traceability: the result is related to a reference through an unbroken documented chain of calibrations, each contributing uncertainty. Traceability does not guarantee that a sensor is fit for purpose; it only supports the credibility of the comparison.

6. Actuators and Output Interfaces

Actuators include relays, contactors, solid-state switches, motors, solenoid valves, heaters, lamps, and variable-speed drives. Selection depends on load voltage and current, inrush current, switching frequency, duty cycle, thermal dissipation, fail-safe behaviour, noise, and required lifetime.

A microcontroller must not normally drive a relay coil or motor directly. A transistor or MOSFET driver, flyback diode for a DC inductive coil, gate resistor where appropriate, current limiting, and separate power path are typical requirements. AC loads may require snubbers, zero-cross switching, or properly rated solid-state devices. Contactors are used for higher power but require consideration of coil suppression and contact wear.

Actuator commands should specify the safe state during power loss, communication failure, watchdog reset, invalid sensor data, and emergency stop. For heating, pumping, and motion systems, software interlocks should supplement—not replace—appropriate hardware protection.

Pulse-width modulation controls average power in many loads. For an ideal resistive load supplied by a switching waveform, the average power is approximately related to duty ratio D by:

This approximation assumes the load and switching frequency make the relevant physical response proportional to average power. It is not universally valid for motors, converters, lamps, or nonlinear loads.

7. Sensor Selection and Interface Design

A disciplined selection process is more reliable than choosing the cheapest or most familiar sensor.

1. Define the measurand and required decision. Is the system measuring true energy, detecting a fault, controlling temperature, or merely indicating status?
2. Specify range, overload, accuracy, resolution, bandwidth, response time, environment, and lifetime.
3. Determine whether isolation and certification are mandatory.
4. Select the physical sensing principle and interface: analog voltage, current loop, frequency, pulse, I²C, SPI, UART, CAN, or industrial protocol.
5. Check loading, power consumption, cable length, electromagnetic compatibility, and grounding.
6. Design conditioning, protection, sampling, and calibration.
7. Verify the complete assembled system against a traceable reference.
8. Plan maintenance, sensor replacement, firmware updates, diagnostics, and end-of-life handling.

Digital output does not automatically mean accurate output. A digital sensor can be poorly calibrated, badly placed, or affected by communication errors. Conversely, an analog sensor can provide excellent performance if its reference, conditioning, and conversion are well designed.

8. Illustrative Case: Smart Motor Monitoring in a Commercial Building

Background

A commercial building operates several induction-motor-driven air-handling units. The facilities team wants early warning of overload, bearing degradation, abnormal vibration, and inefficient operation without installing unsafe temporary instruments.

Problem

The existing controller provides only start/stop status. The proposed IoT node must estimate electrical power, detect vibration changes, monitor motor temperature, and report alarms. The system must tolerate electrical noise and continue safe operation if the network fails.

Analysis

A split-core CT is selected for phase-current measurement because it provides isolation and can be installed without disconnecting the conductor. An isolated voltage transducer measures line voltage. Because the motor is a three-phase load, each phase is sampled where practical; using only one phase would hide unbalance. A three-axis MEMS accelerometer is mounted firmly near the drive-end bearing. A surface-mounted temperature sensor measures bearing-housing temperature, while a separate ambient sensor supports thermal interpretation.

The current and voltage channels are simultaneously sampled. Instantaneous products are averaged over successive electrical cycles to estimate real power. The vibration channel is sampled at a rate high enough for the selected diagnostic band, with an anti-aliasing filter. A watchdog, local alarm threshold, and nonvolatile event buffer provide resilience during network outages.

Approach and solution

The input circuits include fusing, surge protection, isolation, shield and grounding provisions, and physically separated hazardous and low-voltage areas. The CT burden is permanently secured so that the secondary cannot be left open. Factory calibration is checked at several current and voltage points, and temperature drift is characterized during commissioning.

The software reports raw values, quality flags, timestamps, sensor-health status, and derived quantities. A power alarm is not triggered by one noisy sample; it requires persistence over a defined interval. Vibration indicators are compared with a baseline established after the machine is known to be healthy. Maintenance staff receive a trend and diagnostic context rather than an unexplained binary alarm.

Outcome

The system identifies a gradual rise in vibration accompanied by a small increase in bearing temperature. Inspection finds lubrication deterioration before a catastrophic failure. The energy data also reveals that one motor operates with persistent phase imbalance, prompting investigation of connections and supply quality.

Learning points

The case demonstrates that sensor choice depends on the decision to be made. It also shows the importance of isolation, synchronized sampling, baseline data, sensor placement, uncertainty, local fail-safe behaviour, and human interpretation. An IoT dashboard is valuable only when the underlying measurement chain is credible.

9. Challenges, Ethics, and Contemporary Practice

Sensor drift, ageing, contamination, electromagnetic interference, condensation, connector failure, network loss, and software defects can all produce misleading data. Redundancy, plausibility checks, range checks, rate-of-change limits, and sensor-fault flags improve reliability. However, redundant sensors must not share an unnoticed common failure mode.

Occupancy and environmental monitoring can affect privacy and workplace expectations. Data minimization, purpose limitation, access control, retention limits, encryption, and transparent notice are important. Cameras and identity-linked data require stricter governance than anonymous temperature measurements. Cybersecurity also affects physical safety: compromised actuator commands may damage equipment or endanger people. Secure boot, authenticated updates, network segmentation, least-privilege access, logging, and failure-safe controls are appropriate safeguards.

Artificial intelligence can assist anomaly detection and predictive maintenance, but it does not eliminate metrology. A machine-learning model trained on biased, poorly labelled, or drifting sensor data may generate false alarms or miss faults. Models should be validated against physical measurements, monitored for distribution shift, and supplemented by interpretable engineering limits.

Best practice is to preserve a clear distinction between measured data, calibrated estimates, inferred states, and control decisions. Each should have timestamps, units, quality indicators, and documented assumptions.

Chapter Summary

Sensors and transducers form the input foundation of IoT and smart electrical systems, while actuators provide physical outputs. Voltage measurement commonly uses dividers, transformers, or isolated transducers; current measurement may use shunts, current transformers, Hall sensors, or Rogowski coils. Real power is obtained from the time average of instantaneous voltage-current products, and energy is the integral of power.

Temperature, vibration, light, occupancy, humidity, pressure, and gas sensors differ substantially in physical principle, bandwidth, placement, cross-sensitivity, and maintenance needs. Signal conditioning must address amplification, filtering, protection, linearization, isolation, and ADC range. Sampling must prevent aliasing, and synchronized channels are essential for power measurement.

Calibration compares an instrument with a reference, while metrology adds traceability and uncertainty evaluation. Sensor selection must consider range, accuracy, resolution, environment, safety, loading, communication, lifetime, and the actual decision required. Actuator systems require appropriate drivers, suppression, interlocks, fail-safe states, and protection. Finally, trustworthy smart systems combine sound measurement engineering with cybersecurity, privacy governance, diagnostics, and human oversight.

Key Terms

Sensor: Device that responds to a physical quantity.

Transducer: Device that converts a physical quantity or form of energy into another form.

Actuator: Output transducer that produces a physical action from a command.

Measurand: Quantity intended to be measured.

Sensitivity: Change in output per unit change in input.

Resolution: Smallest distinguishable change in a measurement.

Accuracy: Closeness of a result to an accepted reference value.

Precision: Agreement among repeated measurements.

Drift: Change in sensor output over time or environmental conditions.

Hysteresis: Dependence of output on the direction of input change.

Loading: Measurement-system influence on the circuit being measured.

Isolation: Electrical separation between circuits.

Aliasing: Misrepresentation of frequency caused by inadequate sampling.

Calibration: Comparison with a reference under defined conditions.

Traceability: Documented connection of a result to a reference through calibration chains.

Uncertainty: Quantified doubt associated with a measurement result.

Transimpedance amplifier: Amplifier that converts input current into output voltage.

Power factor: Ratio of real power to apparent power.

Anti-aliasing filter: Filter that limits input frequency content before sampling.

Creepage: Surface distance between conductive parts separated by insulation.

Review Questions

5 Short Answer Questions

1. Distinguish accuracy from precision.
2. Why must a CT secondary not be left open while primary current flows?
3. State two reasons for using electrical isolation in an IoT measurement system.
4. What is the purpose of an anti-aliasing filter?
5. Why is sensor placement important for environmental measurements?

5 Long Answer Questions

1. Explain the complete measurement chain from measurand to cloud-based decision, identifying possible error sources at each stage.
2. Compare shunt, current-transformer, Hall-effect, and Rogowski-coil current sensors.
3. Explain real-power measurement for sinusoidal and distorted waveforms, including the assumptions behind each method.
4. Discuss calibration, traceability, standard uncertainty, and expanded uncertainty.
5. Describe the design requirements for a safe actuator interface controlling an inductive load.

3 Analytical / Application Questions

1. An IoT node must monitor a 230 V single-phase heater drawing up to 12 A and report energy over a month. Propose the voltage and current sensors, isolation method, sampling strategy, calibration plan, and safety controls.
2. A vibration sensor reports increasing RMS acceleration, but the motor temperature and electrical power remain normal. Develop a diagnostic plan that distinguishes sensor fault, mounting error, and genuine mechanical deterioration.
3. Design an occupancy-monitoring system for a university laboratory. Compare PIR, radar, ultrasonic, and camera-based approaches, and justify a solution using privacy, reliability, maintenance, and cybersecurity criteria.

Multiple Choice Questions with Answers

Question 1

Which expression gives instantaneous single-phase real power?

A. p(t) = v(t)i(t)

B. p(t) = v(t) + i(t)

C. p(t) = V_rms/I_rms

D. p(t) = √(v(t)² + i(t)²)

Correct Answer: A. p(t) = v(t)i(t)

Question 2

A 10-bit ADC has a 0–5 V input span. What is its approximate quantization step?

A. 0.49 V

B. 4.88 mV

C. 0.488 mV

D. 10 V

Correct Answer: B. 4.88 mV

Question 3

Which sensor is inherently unsuitable for measuring DC current?

A. Shunt resistor

B. Hall-effect sensor

C. Current transformer

D. Fluxgate current sensor

Correct Answer: C. Current transformer

Question 4

What is the principal purpose of an anti-aliasing filter?

A. Increase sensor supply voltage

B. Limit frequency content before sampling

C. Provide galvanic isolation in every case

D. Convert digital data into torque

Correct Answer: B. Limit frequency content before sampling

Question 5

A load operates at 230 V, 8 A, and power factor 0.85. Its approximate real power is:

A. 156 W

B. 1.56 kW

C. 2.16 kW

D. 18.4 kW

Correct Answer: B. 1.56 kW

Question 6

Which statement about a shunt current sensor is correct?

A. It measures current without any power loss

B. Its output voltage is proportional to current

C. It can measure only alternating current

D. It always provides galvanic isolation

Correct Answer: B. Its output voltage is proportional to current

Question 7

Which quantity is directly associated with the closeness of repeated readings to one another?

A. Precision

B. Range

C. Creepage

D. Sensitivity

Correct Answer: A. Precision

Question 8

For independent uncertainty components, the usual combined standard uncertainty is calculated by:

A. Adding only the largest component

B. Taking the arithmetic mean

C. Taking the root-sum-square

D. Multiplying all components

Correct Answer: C. Taking the root-sum-square

Question 9

Which actuator interface is normally required when a microcontroller controls a relay coil?

A. Direct connection to an ADC input

B. A driver stage and suitable coil suppression

C. A photodiode in series with the coil

D. An unconnected CT burden

Correct Answer: B. A driver stage and suitable coil suppression

Question 10

Why should voltage and current channels be synchronized in digital power measurement?

A. To eliminate the need for calibration

B. To ensure instantaneous products represent the correct phase relationship

C. To increase the mains frequency

D. To make all sensors measure temperature

Correct Answer: B. To ensure instantaneous products represent the correct phase relationship