5.2 Measurements, Instrumentation, Sensors, Circuits, and Uncertainty

Think in measurement chains

A sensor is only one link of a measurement system. A complete chain begins with the physical variable, converts it to a signal (a transducer), conditions that signal (amplify, filter, isolate, linearize), digitizes or displays it, and finally uses the value for a decision. FE Mechanical questions may name a device, but the underlying skill is matching the physical principle to the measured quantity and reasoning about the whole chain.

Common FE sensor mappings:

A strain gauge with gauge factor GF relates fractional resistance change to strain: ΔR/R = GF·ε, and is read in a Wheatstone bridge so a small ΔR produces a measurable output voltage. A thermocouple generates a small EMF from the junction temperature difference (Seebeck effect).

Vocabulary and error: accuracy vs. precision

The FE separates terms that students often blur:

• Accuracy — closeness to the true value (small systematic error/bias).
• Precision / repeatability — closeness of repeated readings to each other (small random scatter).
• Resolution — smallest change the instrument can detect.
• Sensitivity — output change per unit input change (slope of the calibration curve).
• Bias — a constant offset; removed by calibration.
• Linearity / hysteresis — deviation from a straight calibration and path dependence.

A device can be precise but inaccurate (tight cluster, wrong center). Calibration corrects bias but not random scatter. Watch the full-scale (% FS) trap: a gauge rated ±1% FS on a 0–1000 kPa range has ±10 kPa of error everywhere. At a 100 kPa reading that is 10% of reading, not 1%. Always convert a % FS spec to an absolute error using the full-scale value, then to relative error using the actual reading.

A/D conversion: an n-bit converter spanning a range R has resolution R/2ⁿ. A 12-bit ADC over 0–10 V resolves 10/4096 ≈ 2.44 mV per step — a frequent quick-calculation question.

Electricity & Magnetism: the circuit behind the sensor

Because sensors output electrical signals, FE Mechanical includes an Electricity & Magnetism domain. Master the core relations:

• Ohm's law: V = IR. Power: P = VI = I²R = V²/R.
• Series resistors: R_eq = R₁+R₂+…; parallel: 1/R_eq = 1/R₁+1/R₂+….
• Kirchhoff's Current Law (KCL): currents into a node sum to zero. Kirchhoff's Voltage Law (KVL): voltage drops around a loop sum to zero.
• Capacitor: Q = CV, energy = ½CV²; in DC steady state it blocks current (open). Inductor: energy = ½LI²; in DC steady state it passes current (short).
• AC: capacitive reactance X_C = 1/(ωC), inductive reactance X_L = ωL, impedance Z = √(R² + (X_L − X_C)²). Real power P = VI·cos θ, where cos θ is the power factor.

Worked example. A 12 V source drives 4 Ω and 8 Ω resistors in series. Total R = 12 Ω, so I = 12/12 = 1 A. Power in the 8 Ω resistor = I²R = (1)²(8) = 8 W. If those same resistors were in parallel: R_eq = (4·8)/(4+8) = 32/12 = 2.67 Ω, and total current rises to 12/2.67 = 4.5 A. These one-step Ohm/Kirchhoff calculations are the bread-and-butter of the E&M domain.

Uncertainty propagation

Uncertainty must follow the operation, not a generic percentage rule. For independent random errors, combine in quadrature (root-sum-square, RSS):

• Addition/subtraction (R = A ± B): combine absolute uncertainties — δR = √(δA² + δB²).
• Multiplication/division (R = AB or A/B): combine relative uncertainties — (δR/R) = √((δA/A)² + (δB/B)²).
• Power (R = Aⁿ): δR/R = |n|·(δA/A).

Example. Power P = VI with V = 10 V ± 2% and I = 2 A ± 3%. Since P is a product, δP/P = √(0.02² + 0.03²) = √(0.0004 + 0.0009) = √0.0013 = 3.6%. The dominant term controls the result, so improving the worst measurement helps most. This RSS approach—relative for products, absolute for sums—is the standard FE method and replaces naïve linear addition of percentages.

Dynamic response of instruments

Real sensors do not respond instantly, and the FE tests how fast a measurement system tracks a changing input. Most instruments are modeled as first-order or second-order dynamic systems — the same forms used in controls.

A first-order instrument (e.g., a thermometer or thermocouple reaching thermal equilibrium) is governed by a single time constant τ. Its response to a step change is exponential: the reading reaches 63.2% of the final value in one time constant, 86.5% in 2τ, and about 98% in 4τ. A bare-bead thermocouple has a small τ and responds quickly; the same junction inside a heavy thermowell has a large τ and lags. If you sample a fast-changing process with a slow sensor, you record a smoothed, delayed version of the truth — a classic measurement error the FE highlights.

A second-order instrument (e.g., a galvanometer, accelerometer, or pressure transducer with mass and spring) adds a natural frequency ωn and a damping ratio ζ. Underdamped instruments (ζ < 1) overshoot and ring; overdamped instruments (ζ > 1) are sluggish; near-critical damping (ζ ≈ 0.6–0.7) gives the best compromise of fast response with little overshoot.

Sampling matters too. When digitizing, the Nyquist criterion requires the sampling rate to exceed twice the highest frequency of interest; sampling too slowly produces aliasing, where a high-frequency signal masquerades as a false low-frequency one. An anti-aliasing low-pass filter ahead of the A/D converter prevents this. The FE expects you to spot when a slow sensor, an undamped instrument, or too-low a sampling rate corrupts the measurement.