Electronic Measurement, Simulation, and Product Design

Introduction and Context

Electronic engineering depends on three mutually reinforcing activities: measuring physical behaviour, predicting behaviour through simulation, and converting requirements into reliable products. Measurement provides evidence about a real circuit; simulation provides a controlled model of circuit behaviour; product design integrates technical, economic, safety, regulatory, and human considerations. A competent engineer must understand not only how to obtain a numerical result, but also whether the result is valid, traceable, repeatable, and relevant to the design decision.

In basic electronics, measurement may involve voltage, current, resistance, frequency, phase, power, noise, temperature, or waveform distortion. At postgraduate level, the central concern is uncertainty: every instrument alters the system to some degree, every model makes assumptions, and every product specification represents a compromise. Consequently, electronic design is an iterative process rather than a linear sequence.

Figure: Electronic product design is an iterative loop linking requirements, simulation, measurement, and redesign.

Learning Objectives

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

1. Explain the principles, architectures, and limitations of electronic measurement systems.
2. Distinguish accuracy, precision, resolution, sensitivity, linearity, repeatability, and uncertainty.
3. Select appropriate instruments and measurement procedures for analogue and digital circuits.
4. Formulate, simulate, and critically validate circuit models.
5. Apply uncertainty propagation, loading analysis, and frequency-domain reasoning.
6. Develop an electronics product from requirements through verification and release.
7. Evaluate reliability, safety, electromagnetic compatibility, manufacturability, ethics, and regulatory obligations.

1. Measurement as an Engineering Process

A measurement is not merely a displayed number. It is an experimentally obtained estimate of a measurand, together with a statement of uncertainty and the conditions under which the estimate applies. A complete measurement chain normally contains a sensor or probe, signal conditioning, conversion, processing, display or data storage, and a reference system.

The ideal voltmeter has infinite input resistance and zero bandwidth limitation. Real voltmeters have finite resistance, capacitance, common-mode limits, noise, offset, drift, quantisation, and calibration error. The ideal ammeter has zero resistance; a real ammeter introduces burden voltage. These non-idealities can change the circuit being tested.

Important measurement terminology

Accuracy and precision are not interchangeable. A poorly calibrated instrument may produce precise but inaccurate results. A high-resolution display also does not guarantee high accuracy.

Measurement architectures

A digital multimeter commonly uses an attenuator or shunt, an analogue-to-digital converter (ADC), a reference, timing circuitry, and digital processing. An oscilloscope uses an input attenuator, amplifier, ADC, memory, trigger system, and display. A lock-in amplifier extracts a small periodic signal by multiplying the input by a reference waveform and low-pass filtering the result; it is valuable for measurements limited by broadband noise.

For an ideal N-bit ADC with input range V_FS, the nominal quantisation step is:

Here q is the voltage represented by one least-significant-bit interval, V_FS is the full-scale range, and N is the number of bits. The ideal quantisation uncertainty is commonly modelled as ±q/2 for a value rounded to the nearest code. This model assumes an ideal ADC, a uniformly distributed quantisation error, and no offset, gain, noise, missing-code, or nonlinearity errors.

The theoretical signal-to-quantisation-noise ratio for a full-scale sinusoid is approximately:

This is an ideal result and does not represent the effective number of bits (ENOB) of a practical converter.

Loading and probe effects

If a source with Thevenin resistance R_S drives a voltmeter of resistance R_M, the measured voltage is:

V_TH is the open-circuit Thevenin voltage. The expression assumes a linear resistive source and instrument at the frequency of interest. If the meter resistance is much larger than R_S, loading is small; otherwise, the measurement may be substantially lower than the open-circuit value. Oscilloscope probes add capacitance, so high-frequency measurements require compensation and a defined probe ground connection.

Uncertainty and propagation

For independent input quantities x₁, x₂, …, xₙ and a result y = f(x₁, x₂, …, xₙ), the combined standard uncertainty is approximated by:

The terms u(x_i) are standard uncertainties, and the partial derivatives are sensitivity coefficients. This first-order formula assumes small uncertainties, a sufficiently smooth function, and negligible covariance. If inputs are correlated, covariance terms must be added:

For a resistor divider, V_O = V_S R₂/(R₁ + R₂), uncertainty may arise from the source, both resistors, temperature, and loading. Reporting V_O = 2.50 V without uncertainty can imply a false level of knowledge; a result such as (2.50 ± 0.03) V, with a stated coverage convention, is more informative.

Measurement practice

A defensible procedure begins by defining the measurand and operating condition. The engineer then selects an instrument whose range, bandwidth, input impedance, common-mode rating, and uncertainty are suitable. Connections should be inspected for grounding and safety hazards. The instrument should be zeroed and, where required, calibrated. Several readings should be recorded rather than relying on a single value. Finally, raw data, environmental conditions, instrument identifiers, settings, and analysis should be preserved.

For periodic signals, sampling must satisfy the Nyquist condition f_s > 2f_max for a strictly band-limited signal. In practice, the sampling rate is selected higher than this limit and an anti-alias filter is used. Aliasing cannot generally be removed after sampling because different analogue frequencies may produce the same sampled sequence.

2. Simulation and Model Credibility

Simulation is the numerical execution of a model. It is not a substitute for physical evidence. A model may be useful for one purpose and invalid for another: a small-signal amplifier model can predict gain near a bias point but cannot predict transistor breakdown or large-signal clipping.

Levels of electronic models

A hierarchy of models is useful:

• Behavioural models describe input–output relationships without internal physical detail.
• Circuit-level models represent resistors, capacitors, transistors, sources, and parasitic elements.
• Device-physics models describe charge transport, temperature effects, geometry, and material properties.
• System-level models represent interfaces, control algorithms, communication channels, and power flows.

Model selection should reflect the question being asked. Excessive detail increases computation and parameter uncertainty; insufficient detail produces misleading confidence.

Major simulation types

DC operating-point analysis finds steady-state node voltages and branch currents. AC small-signal analysis linearises a nonlinear circuit around an operating point and estimates frequency response. Transient analysis solves time-domain behaviour and is required for switching, startup, modulation, and distortion. Noise analysis estimates the contribution of device and resistor noise. Parameter sweeps expose sensitivity to component values, while Monte Carlo analysis samples component tolerances or uncertain parameters.

For a first-order RC low-pass network, the transfer function is:

where R is resistance in ohms, C is capacitance in farads, and s is complex frequency in radians per second. The pole and −3 dB angular frequency are:

These expressions assume ideal components, negligible source and load effects, and linear operation. At frequencies well above f_p, the magnitude decreases at approximately 20 dB per decade.

Numerical convergence and interpretation

Circuit simulators solve algebraic and differential equations using numerical methods. Convergence problems may result from floating nodes, ideal voltage-source loops, discontinuities, unrealistic component values, or unsuitable time steps. Adding arbitrary numerical elements may force convergence but can invalidate the physical interpretation. A better approach is to identify the physical cause, define realistic parasitics, set appropriate tolerances, and compare limiting cases.

Simulation results should be checked using conservation laws, dimensional analysis, limiting behaviour, and hand estimates. For example, a passive network should not deliver more real power than its source under the defined conditions. A simulated 1 MHz circuit may be unreliable if the model contains no package inductance, interconnect capacitance, or frequency-dependent device data.

Verification and validation

Verification asks, “Was the model or computation implemented correctly?” Validation asks, “Does the model represent the real system adequately for its intended use?” Verification may involve mesh refinement, solver comparisons, unit tests, and analytical solutions. Validation requires measured data collected under comparable conditions. A model is never validated in an absolute sense; it is validated for a purpose, range, and set of assumptions.

Sensitivity analysis determines which parameters most affect an output. For output y and parameter p, a local relative sensitivity can be expressed as:

This dimensionless quantity is meaningful when y and p are nonzero and the local derivative is representative. Large sensitivity indicates that tolerance control, calibration, or redesign may be valuable.

3. From Requirements to Electronic Product

Product design begins with a measurable problem statement, not with a preferred component. Requirements should define function, performance, interfaces, environmental conditions, safety, size, cost, service life, manufacturing volume, and regulatory constraints. Each requirement should be testable and assigned a verification method: inspection, analysis, demonstration, or test.

A useful requirements hierarchy is:

1. User need: the problem experienced by the user.
2. System requirement: measurable system-level performance.
3. Subsystem requirement: allocated performance for hardware, firmware, and mechanical elements.
4. Component specification: limits that permit implementation.
5. Verification criterion: evidence required for acceptance.

For example, “low power” is ambiguous. “The device shall consume no more than 250 mW from a 5 V supply in active mode at 25 °C” is testable, although its duty cycle and tolerance may still need definition.

Architecture and trade-offs

Architecture allocates functions among analogue circuits, digital processing, firmware, sensors, power conversion, communication interfaces, and mechanical packaging. Trade-offs may involve cost versus accuracy, bandwidth versus noise, efficiency versus thermal margin, integration versus repairability, and time to market versus design maturity.

A weighted decision matrix can compare alternatives. If option j receives score r_ij for criterion i with weight w_i, its weighted score is:

Weights should be normalised consistently, and scores must be defined before evaluating alternatives. The matrix supports judgement; it does not eliminate bias or uncertainty.

Design calculations and margins

An analogue front end must consider input range, gain, offset, noise, bandwidth, common-mode range, input impedance, protection, and ADC drive requirements. A power supply must consider nominal output, regulation, ripple, transient response, efficiency, thermal dissipation, isolation, and fault behaviour.

For a linear regulator, an approximate power loss is:

This assumes negligible quiescent current and ignores switching or dynamic effects. The junction temperature may be estimated by:

where T_J is junction temperature, T_A is ambient temperature, and θ_JA is junction-to-ambient thermal resistance in °C/W. The estimate is valid only for the specified mounting, airflow, board area, and steady-state conditions. If the result approaches the semiconductor’s maximum rating, thermal transients, tolerances, and derating must be considered.

Prototype, verification, and release

A prototype should be designed for measurement. Test points, accessible grounds, configuration links, current-monitoring locations, and diagnostic firmware reduce debugging time. Bring-up should proceed from visual inspection and continuity checks to power-off resistance tests, current-limited power application, rail verification, clock verification, and staged functional tests.

Design verification demonstrates compliance with requirements. Design validation demonstrates that the product meets user needs in its intended environment. Production tests should be repeatable, fast enough for manufacturing, and linked to failure modes. A design-for-test strategy may include boundary scan, built-in self-test, calibration data, and automated fixture testing.

4. Reliability, Safety, EMC, and Governance

Reliability is the probability that a product performs its required function for a stated time under stated conditions. For a constant failure rate λ, the exponential reliability model is:

Here t is time and λ is the constant failure rate. The model assumes statistically independent failures and a constant hazard rate; it does not describe infant mortality or wear-out regions well. Real products require component quality data, thermal analysis, stress derating, environmental testing, and failure-mode analysis.

Failure Mode and Effects Analysis (FMEA) identifies failure modes, causes, effects, controls, and corrective actions. Risk priority scoring can help organise work, but numerical rankings should not conceal severe hazards or compensate for inadequate engineering judgement.

Electrical safety includes shock, fire, overvoltage, stored energy, isolation, creepage, clearance, fusing, grounding, and enclosure design. Measurement equipment must be rated for the circuit category and voltage. Never connect an oscilloscope ground clip to a floating high-side node unless the instrument and isolation arrangement explicitly permit it.

Electromagnetic compatibility includes both emissions and immunity. Layout practices include short return paths, controlled current loops, suitable decoupling, separation of noisy and sensitive circuits, shielding where appropriate, and carefully designed cable interfaces. Compliance testing is not a substitute for EMC-conscious architecture.

Product development also raises governance issues. Engineers must maintain traceable requirements, version-controlled design files, calibration records, software provenance, and change approval. Safety-critical or connected products require attention to cybersecurity, privacy, update mechanisms, supply-chain risk, and responsible disclosure. AI-assisted design tools can accelerate schematic generation, code production, optimisation, and anomaly detection, but generated results require independent verification because training data, hidden assumptions, and plausible-looking errors can compromise safety.

Worked Example: Measurement Loading and Uncertainty

A source has an open-circuit voltage V_TH = 5.00 V and source resistance R_S = 100 kΩ. A voltmeter has input resistance R_M = 1.00 MΩ. The indicated voltage is:

Thus, the instrument causes an error of approximately 0.455 V relative to the open-circuit value, or 9.09%. A meter with 10 MΩ input resistance would give:

where resistances are expressed consistently in megaohms. The second instrument substantially reduces loading, but probe capacitance and bandwidth may still matter in a time-varying circuit.

Applied Case: Illustrative Low-Power Environmental Monitor

Background

A design team is developing a battery-powered monitor that measures temperature and humidity, stores data, and transmits a packet every ten minutes. The target operating life is one year, and the device must operate from −10 °C to 50 °C.

Problem

Initial simulation predicted adequate battery life, but a prototype failed after several weeks. Measurements also disagreed with the simulator by approximately 8% during radio transmission.

Analysis

The first investigation defined the measurand: average battery current over a complete ten-minute cycle, not the instantaneous current shown by a handheld meter. A shunt resistor and differential data-acquisition channel were selected with sufficient bandwidth. The team measured sleep, sensor conversion, processor activity, and radio burst separately.

The discrepancy arose from three omissions. The simulation used nominal component values, excluded regulator quiescent current, and modelled the radio as a constant-current load rather than a pulsed load. The prototype also had a regulator dropout condition at low battery voltage. Temperature testing showed that battery internal resistance increased at low temperature, producing a larger voltage sag during transmission.

Approach and solution

The team updated the model with regulator efficiency, quiescent current, pulsed radio demand, battery resistance versus temperature, and component tolerances. A Monte Carlo analysis estimated the distribution of operating life rather than reporting one optimistic value. Firmware was modified to shorten the active interval, and the power supply was redesigned with adequate transient capacitance and voltage margin. Production test points were added for battery current and radio supply voltage.

Outcome

The revised prototype met the communication requirement across the temperature range and achieved the specified life with a defined design margin. More importantly, the team established agreement between simulation and measurement by comparing the same operating conditions and by documenting model limitations.

Learning points

The case demonstrates that average performance cannot be inferred from a nominal simulation, that pulsed loads require time-resolved measurement, and that environmental conditions belong in both the model and the verification plan. It also shows why a product requirement should specify test conditions, duty cycle, tolerances, and acceptance limits.

Contemporary Developments and Best Practices

Modern instruments increasingly combine high-speed ADCs, embedded processors, connectivity, and automated analysis. Digital twins and hardware-in-the-loop systems allow control software to be tested against simulated plants before hardware is complete. Machine learning can classify waveforms or identify abnormal signatures, but it should be supported by calibrated data, uncertainty estimates, and a known operating domain.

Best practice is to:

• define the measurand, model purpose, and acceptance criterion before testing;
• select instruments by bandwidth, loading, safety category, uncertainty, and calibration status;
• preserve raw data and metadata;
• use analytical estimates to challenge simulation results;
• include parasitics, tolerances, temperature, and ageing when they affect decisions;
• validate models against independent measurements;
• design prototypes for access, diagnosis, and repeatable testing;
• maintain traceability from requirement to verification evidence;
• apply safety, EMC, cybersecurity, and environmental considerations from the architecture stage;
• communicate uncertainty rather than presenting unsupported precision.

Challenges and Limitations

Measurement is limited by instrument noise, calibration drift, bandwidth, loading, grounding, environmental variation, and operator technique. Simulation is limited by model fidelity, parameter uncertainty, solver assumptions, convergence, and the danger of mistaking numerical precision for physical accuracy. Product design is limited by incomplete requirements, supply-chain changes, manufacturing variation, certification cost, and conflicting stakeholder priorities.

No single model or instrument is universally best. A robust engineering conclusion is one that states its assumptions, identifies dominant uncertainties, survives independent checks, and remains useful for the decision being made.

Chapter Summary

Electronic measurement, simulation, and product design form an iterative engineering discipline. Measurement produces evidence but is affected by loading, bandwidth, calibration, noise, sampling, and uncertainty. Simulation provides prediction through models whose credibility depends on appropriate abstraction, numerical verification, and validation against physical data. Product design translates user needs into testable requirements, architecture, prototypes, verification, validation, manufacturing, and lifecycle support.

Important quantitative tools include ADC quantisation relations, Thevenin loading analysis, uncertainty propagation, RC frequency response, sensitivity analysis, thermal estimates, weighted trade-off scoring, and reliability modelling. Their assumptions must always be stated. Professional practice further requires attention to safety, EMC, reliability, traceability, cybersecurity, ethics, and responsible use of AI-assisted tools. The mature engineer does not ask whether a result is merely precise; the engineer asks whether it is accurate enough, uncertainty-aware, reproducible, and relevant to the intended decision.

Key Terms

Measurement: Experimental determination of a measurand.

Measurand: Quantity intended to be measured.

Accuracy: Closeness to an accepted reference value.

Precision: Closeness among repeated results.

Resolution: Smallest detectable increment.

Sensitivity: Output change per unit input change.

Loading: Disturbance caused by an instrument or test connection.

Traceability: Link from calibration to recognised standards.

Uncertainty: Quantified doubt associated with a result.

Quantisation: Mapping of continuous amplitude to discrete ADC codes.

Aliasing: Misrepresentation of frequency caused by inadequate sampling.

Behavioural model: Model describing system behaviour without internal detail.

Operating point: Steady-state condition about which a circuit may be linearised.

Monte Carlo analysis: Repeated simulation using sampled uncertain parameters.

Verification: Demonstration that specified requirements are met.

Validation: Demonstration that the product fulfils its intended purpose.

FMEA: Structured analysis of failure modes, effects, and controls.

Electromagnetic compatibility: Ability to operate without unacceptable electromagnetic interference.

Derating: Operating below a component’s maximum rating to improve reliability.

Digital twin: Digital representation linked to a physical system or process.

Review Questions

Short Answer Questions

1. Distinguish accuracy from precision and give one example of each.
2. Why does an oscilloscope probe affect high-frequency measurements?
3. State the difference between verification and validation.
4. What is aliasing, and how can it be reduced?
5. Why should a simulation model include tolerances and parasitics when appropriate?

Long Answer Questions

1. Explain the architecture of a digital measurement system and discuss its principal error sources.
2. Derive the transfer function and cutoff frequency of an ideal RC low-pass circuit, stating all assumptions.
3. Discuss model credibility, including verification, validation, sensitivity analysis, and uncertainty.
4. Describe a complete electronics product-development process from requirements through production test.
5. Explain how safety, EMC, reliability, and cybersecurity influence electronic product architecture.

Analytical / Application Questions

1. A 2 V source has a 200 kΩ source resistance and is measured by a 1 MΩ voltmeter. Calculate the indicated voltage and percentage loading error relative to the open-circuit voltage.
2. Design a measurement plan for a pulsed battery-powered transmitter. Specify the measurand, sensor or instrument, bandwidth, sampling strategy, uncertainty sources, and acceptance criterion.
3. A simulated amplifier gain agrees with a prototype at 1 kHz but differs substantially at 10 MHz. Propose a systematic investigation involving model assumptions, measurement setup, parasitics, calibration, and validation data.

Multiple Choice Questions with Answers

Question 1

Which property describes closeness of repeated measurements to one another?

A. Accuracy

B. Precision

C. Sensitivity

D. Traceability

Correct Answer: B. Precision

Question 2

An ideal N-bit ADC with full-scale range V_FS has a nominal quantisation step of:

A. V_FS × 2^N

B. N/V_FS

C. V_FS/2^N

D. V_FS/N^2

Correct Answer: C. V_FS/2^N

Question 3

A 100 kΩ source is measured by a 1 MΩ voltmeter. The measured voltage is what fraction of the open-circuit voltage?

A. 0.091

B. 0.500

C. 0.909

D. 1.100

Correct Answer: C. 0.909

Question 4

Which simulation analysis is most appropriate for startup and switching waveforms?

A. DC operating point

B. AC small-signal

C. Transient

D. Static Monte Carlo only

Correct Answer: C. Transient

Question 5

For an ideal RC low-pass circuit, the cutoff frequency is:

A. 2πRC

B. 1/(2πRC)

C. R/(2πC)

D. C/(2πR)

Correct Answer: B. 1/(2πRC)

Question 6

Validation primarily asks which question?

A. Was the code formatted correctly?

B. Does the model or product represent and satisfy its intended purpose?

C. Was the instrument display readable?

D. Was the component purchased cheaply?

Correct Answer: B. Does the model or product represent and satisfy its intended purpose?

Question 7

Which practice directly reduces aliasing before analogue-to-digital conversion?

A. Increasing display brightness

B. Using an anti-alias filter

C. Removing all grounding

D. Reducing the ADC resolution

Correct Answer: B. Using an anti-alias filter

Question 8

For a linear regulator, approximate dissipation is given by:

A. (V_IN − V_OUT)I_OUT

B. V_OUT/(V_IN I_OUT)

C. I_OUT/(V_IN − V_OUT)

D. V_IN + V_OUT + I_OUT

Correct Answer: A. (V_IN − V_OUT)I_OUT

Question 9

Which activity most directly establishes traceability?

A. Changing a resistor without recording it

B. Linking calibration records to recognised standards

C. Using more decimal places on a display

D. Replacing measured data with simulated data

Correct Answer: B. Linking calibration records to recognised standards

Question 10

A prototype measurement disagrees with simulation. What is the most appropriate first response?

A. Delete the measurement

B. Assume the simulator is always correct

C. Compare assumptions, operating conditions, instrument effects, and model fidelity

D. Increase numerical precision only

Correct Answer: C. Compare assumptions, operating conditions, instrument effects, and model fidelity