Electrical Measurement, Instrumentation and Data Acquisition

Introduction and Context

Measurement is the process of comparing an unknown quantity with an accepted standard. Instrumentation is the broader engineering discipline concerned with sensing, converting, transmitting, displaying, recording, and controlling measured variables. In mechatronics, measurement forms the link between a physical system and its controller: sensors observe position, force, temperature, pressure, speed, voltage, and current; signal-conditioning circuits make those observations usable; and data-acquisition systems convert them into digital information.

A measurement is never perfectly exact. Every instrument has limitations arising from calibration, resolution, loading, environmental effects, electrical noise, and the finite quality of the sensor. A competent engineer therefore reports not only a numerical value but also its uncertainty, units, measurement conditions, and confidence in the result.

This chapter develops the principles of analog and digital instruments, bridges, sensors, oscilloscopes, signal conditioning, and data-acquisition systems. It emphasizes practical selection and critical evaluation rather than merely reading instrument displays.

Learning Objectives

After studying this chapter, a student should be able to:

1. Explain the operating principles and applications of common analog and digital electrical instruments.
2. Distinguish accuracy, precision, sensitivity, resolution, repeatability, and uncertainty.
3. Calculate measurement error, uncertainty, bridge output, loading error, and sampling requirements.
4. Select suitable sensors, bridges, oscilloscopes, and data-acquisition hardware.
5. Identify the effects of noise, grounding, bandwidth, aliasing, and quantization.
6. Design and evaluate a basic measurement chain for a mechatronic system.
7. Apply calibration, documentation, safety, and data-integrity practices.

1. The Measurement System

A practical measurement system usually contains the following stages:

Measurand → sensor/transducer → signal conditioning → transmission or isolation → ADC/data acquisition → processing/display/control

The measurand is the physical quantity being measured. A sensor responds directly to the measurand, while a transducer converts one form of energy or information into another. For example, a thermocouple converts a temperature difference into a small voltage, and a strain gauge converts mechanical strain into a resistance change.

Signal conditioning may include amplification, attenuation, filtering, linearization, excitation, bridge completion, isolation, and protection. The analog-to-digital converter (ADC) samples the conditioned signal and represents it as numerical codes. Software can then calculate derived quantities, detect faults, or close a control loop.

Important static characteristics include:

• Range: interval between the minimum and maximum measurable values.
• Span: upper range value minus lower range value.
• Sensitivity: change in output per unit change in input.
• Resolution: smallest input change that can be distinguished.
• Accuracy: closeness of a result to the accepted reference value.
• Precision: closeness among repeated results.
• Repeatability: precision under the same conditions over a short period.
• Hysteresis: different output values for the same input depending on whether the input is increasing or decreasing.
• Linearity: closeness of the calibration curve to a specified straight-line relationship.
• Drift: gradual change in indication when the measurand is constant.

Accuracy and precision are distinct. A poorly calibrated instrument may give tightly clustered but biased results: high precision, low accuracy. Random noise, by contrast, may reduce precision while the average remains close to the reference.

2. Analog and Digital Instruments

2.1 Analog indicating instruments

An analog instrument presents the measured value through pointer deflection over a graduated scale. A permanent-magnet moving-coil (PMMC) instrument produces torque proportional to current in a magnetic field. With a suitable spring, pointer deflection is approximately proportional to current. PMMC instruments are inherently DC instruments; AC measurement requires rectification or another conversion method.

Moving-iron instruments operate on the force or torque produced by a magnetic field acting on iron. Their deflection is approximately related to the square of current, so they can measure both AC and DC, although their scales are often nonuniform. Electrodynamometer instruments use interacting fixed and moving coils and can measure AC or DC power, especially in wattmeters.

Analog meters are useful for observing trends and fluctuations and may have simple, robust construction. Their disadvantages include parallax, limited resolution, mechanical wear, nonlinear scales, and susceptibility to vibration.

2.2 Digital multimeters

A digital multimeter (DMM) commonly measures voltage, current, resistance, continuity, frequency, and sometimes capacitance or temperature. A voltage range normally uses a high-value input resistor, an amplifier or attenuator, an ADC, and a numerical display. A current range routes current through a precision shunt resistor and measures the resulting voltage.

DMM specifications commonly state accuracy as a combination of a percentage of reading and a number of counts:

Here, E_max is the maximum stated error, p is the percentage expressed as a decimal, V_reading is the displayed value, n is the number of counts, and V_count is the value of the least significant displayed digit. This specification applies under stated conditions, including temperature, warm-up time, frequency, and measurement range.

Worked example. A DMM specification is ±(0.5% of reading + 2 counts) on a 20.00 V range. If the display reads 12.34 V, the count value is 0.01 V. Therefore,

The result should be reported approximately as 12.34 V ± 0.08 V, subject to the manufacturer’s conditions. It is not justified to report more certainty merely because the display shows four digits.

2.3 Ammeter and voltmeter loading

An ideal voltmeter has infinite input resistance and draws no current. A real voltmeter has finite resistance R_V, so connecting it changes the circuit. An ideal ammeter has zero resistance. A real ammeter has a burden resistance, which produces a voltage drop and may alter the current.

Consider a source with open-circuit voltage V_s and Thevenin resistance R_s. Connecting a voltmeter gives:

The fractional loading error relative to the open-circuit voltage is:

The formula assumes a linear Thevenin-equivalent source and a resistive voltmeter input. Loading becomes small when R_V is much greater than R_s.

Example. If V_s = 10 V, R_s = 100 kΩ, and R_V = 1 MΩ,

Thus, the measurement is about 9.1% below the open-circuit value. A 10 MΩ meter would reduce the error substantially.

3. Accuracy, Error, and Uncertainty

The true value is an ideal concept; in practice, a reference value or best estimate is used. Measurement error is commonly expressed as:

where x_m is the measured value and x_ref is the reference value. Relative error and percentage error are:

A systematic error has a reproducible bias, such as zero offset or calibration error. A random error varies unpredictably between observations. Repeating measurements can reduce the effect of random variation but does not remove systematic bias.

For repeated observations x₁, x₂, …, x_N, the arithmetic mean is:

The sample standard deviation is:

The standard uncertainty of the mean, assuming independent observations with similar variance, is:

The use of N − 1 makes s an unbiased estimate of population standard deviation under common statistical assumptions. Repetition does not account for calibration uncertainty, resolution, or environmental effects; these must be evaluated separately.

If independent uncertainty components u₁, u₂, …, u_k contribute to a result, a first-order combined standard uncertainty is often estimated by root-sum-square:

This assumes the components are independent or uncorrelated and that linearization is reasonable. Correlated components require covariance terms. An expanded uncertainty may be stated as U = k u_c, where k is a coverage factor selected for the desired confidence interpretation and statistical model.

Calibration

Calibration compares an instrument with a reference standard over a defined range. A calibration record should identify the instrument, reference equipment, environmental conditions, input values, observed outputs, corrections, uncertainty, date, operator, and traceability. Calibration may reveal zero error, gain error, nonlinearity, hysteresis, and drift.

A two-point linear correction can be represented as:

where m is the measured slope and b is the measured intercept of indicated output against reference input. This correction is valid only over the range where the linear model is appropriate. Calibration intervals should reflect risk, usage, stability, manufacturer guidance, and historical drift rather than an arbitrary calendar rule.

4. Sensors and Transducers

Sensors may be classified by the quantity measured, operating principle, output form, or need for external excitation.

Sensor selection should consider range, sensitivity, bandwidth, resolution, environmental compatibility, response time, linearity, hysteresis, reliability, power consumption, cost, and the consequences of failure.

Resistive sensors and Wheatstone bridges

A Wheatstone bridge consists of four resistive arms. With supply voltage V_ex, output measured between the midpoint nodes, and arms R₁, R₂, R₃, R₄, one common polarity convention gives:

The sign depends on which midpoint is defined as positive. The bridge is balanced when the two divider ratios are equal:

For a quarter-bridge strain-gauge arrangement with one active gauge, nominal resistance R, and small fractional resistance change ∆R/R, the magnitude of output is approximately:

This approximation assumes small changes, matched nominal resistances, and an ideal high-impedance measurement input. Half-bridge and full-bridge arrangements can increase sensitivity and provide temperature compensation.

Example. A full bridge has V_ex = 5 V and produces 10 mV. If the small-signal sensitivity is approximately V_o ≈ V_ex(∆R/R), then:

The exact relationship depends on the bridge configuration and which resistors change; the stated approximation must not be transferred indiscriminately between quarter-, half-, and full-bridge circuits.

Instrumentation amplifiers are widely used with bridges because they provide high input impedance, differential measurement, gain, and good common-mode rejection. Their gain must be selected so the largest expected signal remains within the amplifier and ADC input range.

5. Oscilloscopes and Time-Domain Measurement

An oscilloscope displays voltage as a function of time. Major controls include volts per division, seconds per division, trigger level and slope, coupling, probe attenuation, bandwidth, and acquisition mode.

A periodic waveform’s frequency and period are related by:

For a sinusoid, root-mean-square voltage is:

This relation applies to a zero-mean sinusoid, not to arbitrary waveforms. For a general periodic waveform, RMS is defined by:

Probe selection matters. A ×10 passive probe increases effective input resistance and reduces capacitive loading compared with a ×1 probe, but it attenuates the signal and requires correct scale settings. Probe ground leads can form loops that pick up magnetic interference; short spring-ground connections are preferable for fast signals. Differential or isolated probes are required when neither measurement point is safely referenced to instrument ground.

Oscilloscope measurements should specify coupling and bandwidth. AC coupling removes the DC component through a high-pass network; it is useful for ripple but can conceal the absolute operating level. Bandwidth limiting can reduce high-frequency noise, but it also removes legitimate signal content.

6. Noise, Interference, and Signal Integrity

Noise is an unwanted random or deterministic component superimposed on a signal. Common sources include thermal noise, shot noise, electromagnetic interference, ground loops, switching converters, motor commutation, inadequate shielding, and ADC quantization.

Thermal noise voltage for a resistor can be characterized by its mean-square value over bandwidth B:

Here, k is Boltzmann’s constant in J/K, T is absolute temperature in kelvin, R is resistance in ohms, and B is bandwidth in hertz. The expression assumes an ideal resistor and a frequency range in which the model is applicable.

Good practice includes twisted pairs for differential signals, shielded cables where appropriate, single-point signal references, physical separation from power wiring, low-noise power supplies, proper termination, guarding of high-impedance nodes, and analog filtering before conversion. Shielding is not a substitute for correct grounding; a shield connected incorrectly can create a ground-loop current.

Common-mode rejection ratio (CMRR) describes an amplifier’s ability to reject signals common to both inputs. In decibels:

where A_d is differential gain and A_c is common-mode gain. High CMRR is valuable for small bridge signals in electrically noisy environments, but it can be degraded by resistor mismatch, frequency, cable imbalance, and excessive common-mode voltage.

7. Data Acquisition and Digital Conversion

A data-acquisition (DAQ) system typically includes sensors, excitation, analog signal conditioning, multiplexers, sample-and-hold circuits, ADCs, digital interfaces, software, and storage. Design begins with the required measurement bandwidth, accuracy, sample rate, channel count, isolation, synchronization, and environmental rating.

Sampling and aliasing

If a continuous signal contains frequency components above half the sampling frequency, those components can appear falsely at lower frequencies. The Nyquist condition for a band-limited signal is:

where f_s is sampling frequency and f_max is the highest signal frequency of interest. In practice, sampling exactly at twice the maximum frequency is insufficient in the presence of noise, filter roll-off, and frequency uncertainty. An analog anti-aliasing low-pass filter must be placed before the ADC.

The aliased frequency can be found by folding a frequency f around integer multiples of f_s; one useful representation is:

where integer k is chosen to place the result in the baseband interval. Sampling does not recover information already lost to aliasing.

Quantization and ADC resolution

An ideal N-bit ADC with input span V_FS has 2^N codes. The nominal least-significant-bit size is:

Quantization error in an ideal, properly scaled converter is commonly modeled as lying between approximately −q/2 and +q/2, with assumptions about quantization and signal distribution. Real ADC accuracy also includes offset error, gain error, integral nonlinearity, differential nonlinearity, reference error, noise, and temperature drift.

Example. A 12-bit ADC measures 0–5 V. Its nominal step is:

A sensor signal of only 2 mV may therefore be represented by roughly one or two codes before noise and analog errors are considered. Gain should be used where appropriate, while ensuring that expected peaks do not saturate the ADC.

Multiplexing and synchronization

A multiplexed DAQ shares one ADC among multiple channels. The effective time between samples on a particular channel is longer than the overall conversion period. Channel switching can also introduce settling errors, especially with high source impedance. Simultaneous-sampling ADCs are preferred when phase relationships among channels are important, such as power measurement, vibration analysis, and motor control.

Timestamp quality matters in software analysis. A nominal sample rate is not enough if samples are dropped, delayed, or irregularly spaced. Systems used for control should have deterministic timing, bounded latency, and a defined response to communication failure.

8. Worked Measurement-Chain Design

Suppose a mechatronic actuator produces a force from 0 to 800 N. A load cell has a full-scale bridge output of 2 mV/V and is excited at 5 V. The DAQ accepts 0–2.5 V, and the designer wants the full force range to use approximately 2.0 V after amplification.

The load-cell full-scale differential output is:

The required nominal gain is:

A practical instrumentation amplifier would be selected with gain near 200, but its input offset, noise, common-mode range, bandwidth, and output swing must be checked. If a 12-bit, 0–2.5 V ADC is used, its nominal step is:

At a 2.0 V signal span, the ideal code resolution referred to force is approximately:

This is only an ideal quantization estimate. Load-cell nonlinearity, creep, hysteresis, amplifier noise, excitation stability, mechanical mounting, and calibration may dominate the actual uncertainty. If the force changes significantly only below 20 Hz, a sample rate of 200 samples/s with a carefully designed anti-alias filter may be adequate; a vibration measurement would require a substantially higher rate.

9. Applied Case: Force Monitoring in a Robotic Gripper

This is an illustrative engineering case.

Background

A robotic gripper must hold plastic components without crushing them. The existing system commands motor position but has no direct force feedback. Operators report occasional part damage and uncertain grip reliability.

Problem

The engineering team must measure gripping force from 0 to 300 N with a target accuracy better than ±3 N over a factory temperature range. The signal must be available to a controller at 100 Hz, and electrical noise from the motor drive is expected.

Analysis

A strain-gauge load cell is selected because it measures force directly and can be mechanically integrated into the gripper jaw. A full-bridge sensor is preferred over a quarter bridge because it provides greater sensitivity and improved rejection of uniform temperature changes. The bridge output is only a few millivolts, so a high-CMRR instrumentation amplifier is required.

The mechanical structure is analyzed to ensure that off-axis loads do not exceed the load-cell rating. Cable routing is separated from motor phases, and a shielded twisted pair carries the differential bridge signal. The excitation supply is regulated and monitored because bridge output is proportional to excitation voltage.

Approach and solution

The signal chain consists of the full-bridge load cell, low-noise excitation, instrumentation amplifier, second-order low-pass filter, isolated 16-bit DAQ input, and controller software. The analog cutoff is selected above the 100 Hz control bandwidth but below half the sampling rate, allowing a transition band for attenuation. The sensor is calibrated at several increasing and decreasing force points to characterize linearity and hysteresis.

Software applies the calibration slope and offset, checks for open-circuit and out-of-range values, and rejects implausible rate-of-change events. A watchdog causes the gripper to reduce force if data are missing or inconsistent. Calibration coefficients are stored with sensor identification and date information.

Outcome

The system provides closed-loop force control and detects contact earlier than position-only control. Damage is reduced, but the test reveals that mounting friction produces a direction-dependent error. The final design therefore includes a mechanical guide improvement and separate validation under realistic loading directions.

Learning points

1. Sensor accuracy cannot compensate for poor mechanical design.
2. A bridge signal requires attention to excitation, common-mode range, gain, filtering, and grounding.
3. Calibration must represent the actual direction, range, temperature, and mounting conditions.
4. Data validation and fail-safe behavior are part of instrumentation engineering, not optional software features.

10. Safety, Governance, and Professional Practice

Measurement equipment can expose users to hazardous voltages, stored energy, rotating machinery, hot surfaces, and unexpected actuator motion. Instrument category ratings, maximum input specifications, fuses, isolation barriers, protective earth, and safe probing procedures must be respected. A general-purpose oscilloscope ground clip may be connected to protective earth; attaching it to a floating high-voltage node can cause a short circuit or electric shock.

Measurement data should be traceable and protected against silent alteration when used for compliance, safety, quality assurance, or research. Records should retain raw data, calibration status, timestamps, software versions, configuration, uncertainty estimates, and operator identity where appropriate. Engineers should not delete inconvenient observations or report excessive significant figures.

In automated systems, sensor faults can create unsafe decisions. Plausibility checks, redundancy for critical variables, range limits, diagnostic excitation, and independent shutdown paths may be necessary. Artificial intelligence can assist with anomaly detection and sensor-fusion models, but it does not remove the need for calibrated sensors, explainable limits, representative training data, and human responsibility for safety decisions.

11. Best-Practice Design Procedure

A disciplined measurement design can follow these steps:

1. Define the measurand, required range, bandwidth, accuracy, response time, and operating environment.
2. Select a sensor whose physical principle and mechanical interface suit the application.
3. Estimate loading, noise, drift, uncertainty, and failure modes before purchasing hardware.
4. Design excitation, amplification, filtering, protection, grounding, and isolation.
5. Select ADC resolution and sample rate from the signal bandwidth and required uncertainty, not from resolution alone.
6. Calibrate over the intended operating range and document the reference standards.
7. Verify the complete chain using known inputs and realistic disturbances.
8. Validate software scaling, units, timestamps, alarms, saturation behavior, and data storage.
9. Reassess performance after installation because wiring, mounting, temperature, and electromagnetic conditions may differ from laboratory conditions.

Chapter Summary

Electrical measurement and instrumentation connect physical systems to human decisions and automatic control. Analog instruments provide direct visual indication, while digital instruments provide numerical measurement, computation, and data storage. Real instruments introduce loading, finite resolution, calibration error, drift, noise, and bandwidth limitations.

Accuracy describes closeness to a reference, whereas precision describes repeatability. Systematic and random errors require different treatments. Repetition can reduce random uncertainty but cannot remove bias. Calibration establishes the relationship between indication and reference and should include traceability and uncertainty.

Sensors convert physical variables into electrical signals. Wheatstone bridges are especially important for resistive sensors such as strain gauges and RTDs. Instrumentation amplifiers provide high input impedance, differential gain, and common-mode rejection for small bridge outputs. Oscilloscopes reveal time-domain behavior but require correct probe grounding, coupling, triggering, and bandwidth.

DAQ design requires attention to anti-alias filtering, sampling frequency, ADC quantization, multiplexing, synchronization, timing, and data integrity. A successful measurement chain is a complete engineered system: sensor, mechanics, excitation, conditioning, conversion, software, calibration, safety, and validation must all be considered together.

Key Terms

Measurand: Physical quantity intended to be measured.

Sensor: Device that responds to a physical quantity.

Transducer: Device that converts energy or information from one form to another.

Sensitivity: Change in output per unit change in input.

Resolution: Smallest distinguishable input change.

Accuracy: Closeness of a measurement to an accepted reference.

Precision: Closeness among repeated measurements.

Calibration: Comparison with a reference standard to establish measurement performance.

Systematic error: Reproducible measurement bias.

Random error: Unpredictable variation between observations.

Loading error: Error caused by the measuring instrument altering the circuit or system.

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

Wheatstone bridge: Four-arm resistive circuit used for sensitive differential measurement.

Instrumentation amplifier: Differential amplifier with high input impedance and controlled gain.

Common-mode rejection ratio: Measure of rejection of signals common to both amplifier inputs.

Aliasing: False low-frequency representation caused by inadequate sampling.

Quantization: Representation of a continuous amplitude by discrete digital codes.

Data acquisition: Process of sampling, converting, processing, and storing measured signals.

Uncertainty: Parameter describing the dispersion or limited knowledge associated with a measurement result.

Review Questions

Short Answer Questions

1. Distinguish accuracy from precision and give one example of each.
2. Why does a real voltmeter produce loading error?
3. State two advantages of a full-bridge strain-gauge arrangement.
4. What is the purpose of an anti-aliasing filter?
5. Name four important characteristics that should be included in a calibration record.

Long Answer Questions

1. Explain the complete signal path from a physical measurand to a digital controller, identifying the function of each stage.
2. Compare PMMC, moving-iron, analog, and digital instruments with respect to operating principle, strengths, and limitations.
3. Explain systematic and random error, and describe how repeated observations and calibration address them.
4. Describe the design of a bridge-based measurement system, including excitation, amplification, filtering, grounding, and ADC selection.
5. Explain how oscilloscope probe characteristics, grounding, coupling, triggering, and bandwidth affect measurement validity.

Analytical / Application Questions

1. A 5 V source has a Thevenin resistance of 50 kΩ and is measured by a 1 MΩ voltmeter. Calculate the indicated voltage and percentage loading error.
2. Design a sampling and filtering strategy for a vibration signal containing useful components up to 1.2 kHz. Justify the sample rate, transition band, and need for analog filtering.
3. A force sensor produces a 4 mV full-scale output and is connected to a 0–2.0 V ADC through an amplifier. Determine the nominal gain, identify two possible sources of uncertainty, and propose one calibration test for each source.

Multiple Choice Questions with Answers

Question 1

Which characteristic describes closeness of repeated measurements to one another?

A. Accuracy

B. Precision

C. Sensitivity

D. Span

Correct Answer: B. Precision

Question 2

A voltmeter has 10 MΩ input resistance and is connected to a source with 100 kΩ Thevenin resistance. What is the approximate loading error?

A. 0.01%

B. 0.99%

C. 9.09%

D. 10.0%

Correct Answer: B. 0.99%

Question 3

A 12-bit ADC with a 0–4.096 V input span has an ideal nominal step size of:

A. 0.1 mV

B. 1 mV

C. 4 mV

D. 12 mV

Correct Answer: B. 1 mV

Question 4

What is the principal purpose of an instrumentation amplifier in a strain-gauge system?

A. To provide low input impedance

B. To amplify differential signals while rejecting common-mode signals

C. To convert DC into mechanical motion

D. To eliminate the need for calibration

Correct Answer: B. To amplify differential signals while rejecting common-mode signals

Question 5

A signal contains useful frequency components up to 2 kHz. Which sampling frequency satisfies the basic Nyquist condition?

A. 1 kHz

B. 2 kHz

C. 3 kHz

D. 5 kHz

Correct Answer: D. 5 kHz

Question 6

For a zero-mean sinusoidal voltage with peak value 10 V, the RMS value is approximately:

A. 5.00 V

B. 7.07 V

C. 10.00 V

D. 14.14 V

Correct Answer: B. 7.07 V

Question 7

Which error is most directly reduced by calibrating an instrument against a traceable reference?

A. Systematic bias

B. Random thermal noise in every observation

C. Aliasing caused by undersampling

D. Mechanical vibration of an unrelated machine

Correct Answer: A. Systematic bias

Question 8

In a balanced Wheatstone bridge, the two voltage-divider ratios are:

A. Equal

B. Opposite in sign only

C. Always zero

D. Proportional to sampling frequency

Correct Answer: A. Equal

Question 9

Why is an oscilloscope ground clip potentially hazardous on a floating high-voltage node?

A. It always increases probe bandwidth

B. It may connect the node directly to protective earth

C. It converts AC into DC

D. It prevents triggering

Correct Answer: B. It may connect the node directly to protective earth

Question 10

A 5 V, 10-bit ADC has an ideal step size closest to:

A. 0.49 mV

B. 4.88 mV

C. 10 mV

D. 48.8 mV

Correct Answer: B. 4.88 mV