Question 2 of 7: Noise, Sensing Circuits and Data Acquisition

Question 2: Noise, Sensing Circuits and Data Acquisition (20 marks)

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) Aliasing errors occur when a signal is sampled at a rate below the Nyquist rate for its highest-frequency content ($f_s<2f_{max}$). The under-sampled high-frequency energy cannot be distinguished, from the sample sequence alone, from a different, lower ("alias") frequency that the same samples would equally be consistent with — it folds down into the baseband as spurious low-frequency content that no amount of processing after the fact can separate back out.

b) An internal standard is a known, fixed quantity of a chemically similar but distinct substance added in identical amount to every sample and every calibration standard before analysis. Because it experiences the same preparation losses, injection-volume variation and instrument-response drift as the analyte itself, the ratio of the analyte's response to the internal standard's response cancels most of that run-to-run variability, giving far better precision than an absolute (uncorrected) response would.

c) Shot noise originates from the fundamentally discrete, quantized nature of charge carriers (or photons) crossing a potential barrier — a p-n junction, a vacuum-tube cathode, a photodetector absorbing individual photons. Because the arrival of each discrete carrier is a statistically independent random (Poisson) event, the resulting current fluctuates about its mean even under perfectly steady, noise-free DC bias conditions; unlike thermal (Johnson) noise it needs a net current flowing through a barrier, not merely a resistor at a finite temperature.

d) A reference electrode maintains a fixed, precisely known half-cell potential (via an internal redox couple in a saturated, known-concentration internal electrolyte, e.g. Ag/AgCl in saturated KCl) regardless of the composition of the external test solution it is placed in. A porous junction (frit) provides ionic contact with the test solution — completing the measurement circuit — while physically isolating the reference electrode's own internal filling solution from being contaminated by, or diluted into, the sample; this fixed potential is the stable baseline against which a separate measuring/indicator electrode's measurand-dependent potential is compared.

e) An isolation amplifier transmits a signal across a stage with no direct galvanic (conductive) connection between its input and output sides — coupling the signal via a transformer, an opto-isolator, or a capacitive link instead — so the sensor-side circuit and the readout-side circuit can sit at very different ground/common-mode potentials, even kilovolts apart, with no DC current path between them. It is used to break ground loops and to protect equipment and personnel from high voltages or fault currents that may appear on the sensor side.

f) With FM, the information rides on the signal's FREQUENCY (or frequency deviation), not its amplitude, so a receiver that only needs to track zero-crossings/frequency is largely immune to amplitude-based noise and attenuation accumulated over a long transmission path — up to a limiting threshold, additive noise on the line simply does not corrupt the frequency the way it directly corrupts an AM signal's amplitude, which IS the carrier of the information there.

g) Noise is an inherent, RANDOM fluctuation with no identifiable external source — thermal (Johnson) noise, shot noise — intrinsic to the physical measurement process itself; it cannot be eliminated by shielding or filtering, only reduced (cooling, bandwidth limiting) or averaged out statistically. Electrical interference is a DETERMINISTIC, externally coupled unwanted signal from an identifiable source (60 Hz mains hum, a nearby motor, a switching supply) via a specific coupling mechanism (capacitive, inductive, conductive/ground-loop, or radiated); because it has an identifiable source and path, it CAN in principle be removed by appropriate shielding, filtering, grounding practice, or physical separation.

h) A real voltmeter with finite input impedance $R_v$ forms a voltage divider with the sensor's own (often unknown) source impedance $R_s$, so the meter reads $V_{true}\cdot R_v/(R_v+R_s)$ — lower than the true open-circuit voltage. Making $R_v$ very much greater than $R_s$ drives that loading error toward zero, so the instrument reads the sensor's true output essentially without disturbing (loading) it.

i) Fluorescence is measured as a signal detected against an essentially DARK background (with proper excitation/emission filtering, almost no light reaches the detector at the emission wavelength except the fluorescence itself), so even a small concentration produces a signal that stands out clearly above a near-zero baseline. Absorbance, by contrast, is inferred as $A=-\log_{10}(I/I_0)$, a SMALL DIFFERENCE between two large, nearly equal signals (incident $I_0$ and transmitted $I$) at low concentration — the same "subtracting two large numbers" precision-loss problem noted in Q1(f) — so absorbance is intrinsically far less sensitive than fluorescence at low concentrations, where $I$ is barely distinguishable from $I_0$.

j) The trade-off is image quality versus radiation dose (and motion blur). X-ray photon detection is itself governed by Poisson (shot-noise-like) statistics, so image signal-to-noise ratio improves only as the square root of the number of photons collected — a longer exposure (or higher tube current) collects more photons and gives a lower-noise, higher-contrast image, but it also increases the ionizing radiation dose delivered to the patient/operator and increases the risk of motion-blur artifacts. The exposure is set to the minimum needed for diagnostically adequate image quality (as low as reasonably achievable), not to maximize image quality alone.