22-Agric-A5 Principles of Instrumentation · May 2016
Nivaar worked solution (AI-drafted; not reviewed by a licensed engineer)
Paper format. 04-Agric-A5 Principles of Instrumentation, National Exams May 2016 — a three-hour open-book exam; any non-communicating calculator is permitted. Questions 1 and 2 are compulsory (20 marks each); candidates then choose any three (3) of Questions 3-7 (20 marks each) for a 100-mark paper. All seven questions are worked here.
Reference texts. E.O. Doebelin, Measurement Systems: Application and Design, 5th ed. (calibration, standards, static/dynamic sensor characteristics, second-order step response, sampling and ADCs); J.P. Bentley, Principles of Measurement Systems, 4th ed. (accuracy vs. precision, error propagation, signal conditioning); P. Horowitz and W. Hill, The Art of Electronics, 3rd ed. (Johnson noise, CMRR, ADC architectures, anti-aliasing, op-amp signal conditioning); J. Fraden, Handbook of Modern Sensors: Physics, Designs, and Applications, 5th ed. (thermistors, thermocouples, capacitive and photo sensors).
Question text not reproduced: the examination questions are © Engineers and Geoscientists BC. Open the official past paper (linked at the top of this page) to read the question, then follow the worked solution below.
a) R2 only reports what fraction of the output's variance is explained by the fitted model — a curve can have R2 very close to 1 while still being off by a large absolute amount at every point, because R2 is a relative, unitless goodness-of-fit measure that says nothing about the physical size of the residuals. The RMS error reports the residuals themselves, in the instrument's own engineering units, so it directly tells the user the typical size of the reading error they should expect — the number that actually matters when deciding whether the instrument is accurate enough for a given application.
b) Noise is the random, unpredictable component of a measured signal — mathematically, the difference between the actual instantaneous output and the true (noise-free) underlying signal, $n(t)=x_{measured}(t)-x_{true}(t)$, characterized statistically (typically by its standard deviation or RMS value) rather than by any deterministic formula, because by definition it cannot be predicted from past values of itself.
c) ‘Zero’ is the output the instrument reads when the input (measurand) is at its reference/minimum value — the offset of the calibration line. ‘Span’ is the total change in output over the full working range of the input — the difference between the output at full-scale input and the output at zero input. Together, zero and span are the two numbers (offset and slope-times-range) that fully define a linear calibration.
d) A three-point calibration is required whenever a single straight line (two points: zero and span) cannot be trusted to represent the instrument's true response — i.e. whenever the calibration curve is suspected to be non-linear. The third point (typically at mid-range) lets the curvature be detected and quantified rather than silently assumed away.
e) A blank is a measurement of a sample that contains none of the substance being measured (e.g. pure solvent with no analyte) — it establishes the instrument's own baseline/background response and its associated noise floor, which must be subtracted from every real reading. It is repeated several times because the blank reading is itself noisy; only a repeated, averaged blank gives a statistically reliable estimate of both the baseline level and the noise it must be distinguished from (which also sets the detection limit, part i).
f) An interference is any unwanted physical or chemical influence — other than the intended measurand — that also produces a response in the sensor (e.g. a temperature sensor that is slightly sensitive to humidity, or a chemical sensor that cross-reacts with a similar compound). A highly sensitive instrument responds strongly to small changes in its intended input, but that same high gain amplifies these unintended inputs just as strongly, so sensitivity and susceptibility to interference rise together.
g) Differentiation is a high-pass operation: for a sinusoidal component of frequency $f$, $d/dt$ multiplies its amplitude by $2\pi f$, so high-frequency content is boosted proportionally to its frequency. Real signal noise is typically broadband (roughly flat or rising with frequency, e.g. Johnson/shot noise), while the genuine signal of interest is usually concentrated at low frequency, so differentiating amplifies the noise far more than the signal, degrading the signal-to-noise ratio of the derivative compared with the original trace.
h) A representative sample is a small portion of a larger population or bulk material selected (or physically prepared, e.g. by homogenizing/quartering) so that its composition or properties statistically match the average of the whole population, within a stated tolerance — without which a measurement on the sample tells you only about that one, potentially biased, portion, not about the bulk quantity it was meant to characterize.
i) The lowest detectable limit (LDL) is set by comparing the signal to the noise/blank fluctuation level, not by the sensor's nominal resolution: it is conventionally the smallest input whose response exceeds the blank/noise standard deviation by a fixed statistical margin (commonly 3σblank). Below that input, a genuine response cannot be reliably distinguished from random baseline noise.
j) Any of: primary (fundamental) standards traceable directly to an SI definition or a fixed physical constant/phenomenon (e.g. the triple point of water); secondary/certified reference standards calibrated against a primary standard and distributed with a traceable uncertainty (e.g. NIST-traceable weights, certified reference materials); and working standards used for routine day-to-day calibration, themselves periodically re-checked against a secondary standard to maintain the traceability chain.
