Skip to content

Transducers and Detectors

CSIR NET Physical Sciences: temperature, strain, pressure and vacuum, magnetic, vibration, optical and particle transducers and detectors, with practice.

  • 20 min read
  • Updated
On this page · 17 sections
  1. Prerequisites
  2. The Hook: Every Measurement Ends as a Voltage
  3. Classifying Transducers
  4. Temperature
  5. Strain, Force and Pressure
  6. Vacuum Gauges
  7. Magnetic Field
  8. Vibration and Acceleration
  9. Optical Detectors
  10. Particle and Radiation Detectors
  11. Memory Tricks
  12. Don’t Confuse
  13. Common Mistakes
  14. Practice
  15. Quick Revision Table
  16. Key Takeaways
  17. What’s Next

Prerequisites

The Hook: Every Measurement Ends as a Voltage

Classifying Transducers

BasisKindExamples
Energy sourceActive (self-generating): produce an electrical signal from the measurand’s own energythermocouple, piezoelectric crystal, photovoltaic cell, moving-coil pickup
Passive (modulating): change a parameter and need external excitationresistance thermometer, thermistor, strain gauge, LVDT, capacitive sensor
Outputanalogue or digital (shaft encoder, frequency output)
Roleprimary sensing element (diaphragm, bimetal) feeding a secondary electrical transducerdiaphragm plus strain gauge in a pressure sensor

Figures of merit common to all: sensitivity (output per unit input), range, linearity, resolution, hysteresis, response time (often a first-order time constant), drift, and noise. Part C distractors often swap sensitivity (slope) with resolution (smallest detectable change, set by noise).

Temperature

Resistance thermometers (RTDs)

A metal’s resistance rises nearly linearly with temperature. For platinum, the standard sensor (Pt100: 100 Ω at 0 °C),

$$R(T) \approx R_0\left(1 + \alpha T\right), \qquad \alpha \approx 0.00385\ /^\circ\text{C}$$

over moderate ranges (a quadratic correction is used for precision work). Platinum is chosen because it is chemically stable, reproducible and has a wide range. Lead resistance adds directly to the reading, so precise work uses four-wire (Kelvin) connection: two leads carry the current, two separate leads sense the voltage and carry almost no current.

Thermistors

Semiconductor (usually metal-oxide) resistors with a large, usually negative temperature coefficient (NTC):

$$R(T) = R_0\exp\left[B\left(\frac{1}{T} - \frac{1}{T_0}\right)\right], \qquad \alpha = \frac{1}{R}\frac{dR}{dT} = -\frac{B}{T^2}$$

with \( T \) in kelvin and \( B \) a few thousand kelvin. For \( B = 3950 \) K at 25 °C, \( \alpha \approx -4.4\% \) per kelvin, more than ten times a platinum RTD. The price is strong nonlinearity and a narrower useful range. A large measuring current causes self-heating, a systematic error for every resistive sensor.

Thermocouples

Two dissimilar metals joined at a measuring junction produce a Seebeck emf that depends on the temperature difference between that junction and the reference junction:

$$V \approx S_{AB}\,(T_{\text{hot}} - T_{\text{ref}})$$

with \( S_{AB} \) of order tens of µV/K for common base-metal pairs. Two practical laws: the law of intermediate metals (inserting a third metal, such as copper meter leads, adds nothing if its two junctions are at the same temperature) and the law of intermediate temperatures (emfs for \( T_1 \to T_2 \) and \( T_2 \to T_3 \) add). Because the output depends on the reference junction, the reference is held at 0 °C (ice bath) or measured separately and corrected: cold-junction compensation. Thermocouples are active, rugged, cheap and wide-range, but their microvolt signals need careful amplification and are vulnerable to pickup.

Semiconductor diode thermometers

At constant forward current, a silicon diode’s forward voltage falls by roughly 2 mV/K near room temperature. Simple and nearly linear; widely used in cryogenics with calibrated curves.

Other temperature sensors

Gas and vapour-pressure thermometers (primary, low temperature), radiation pyrometers (non-contact, Stefan–Boltzmann or Wien), and for very low temperatures calibrated germanium or ruthenium-oxide resistors.

Strain, Force and Pressure

Strain gauges

A wire or foil stretched by strain \( \varepsilon \) changes resistance through both its geometry and its resistivity:

$$\frac{\Delta R}{R} = G\,\varepsilon, \qquad G = 1 + 2\nu_P + \frac{\Delta\rho/\rho}{\varepsilon}$$

where \( \nu_P \) is Poisson’s ratio. Metal foil gauges have \( G \approx 2 \) (mostly geometry). Semiconductor (piezoresistive) gauges have \( G \) of order 100 (dominated by the resistivity change), much more sensitive but strongly temperature-dependent.

Since \( \Delta R/R \) is around \( 10^{-3} \) or less, gauges are read in a Wheatstone bridge with excitation \( V_{ex} \):

BridgeActive armsOutput (small strain)
Quarter1\( V_{ex}G\varepsilon/4 \)
Half2 (one in tension, one in compression)\( V_{ex}G\varepsilon/2 \)
Full4\( V_{ex}G\varepsilon \)

The exact quarter-bridge output is \( V_{ex}x/(4 + 2x) \) with \( x = \Delta R/R \), slightly nonlinear. An unstrained dummy gauge in the adjacent arm, on the same material and at the same temperature, cancels temperature-induced resistance changes. The bridge itself is treated with the other signal-conditioning circuits in Measurement, Signal Conditioning, Impedance Matching and Feedback.

Pressure sensors

A primary element deforms under pressure: a diaphragm, a bellows or a Bourdon tube (a curved tube that straightens as internal pressure rises). The deflection is converted by a strain gauge, a capacitance change (capacitive manometer: diaphragm as one plate of a capacitor, gas-independent and accurate) or an LVDT. Piezoelectric sensors respond to changing pressure, not static pressure.

Displacement: LVDT and capacitive sensors

The linear variable differential transformer has one primary and two secondary coils wound in series opposition around a movable core. At the centre the secondary voltages cancel; displacement unbalances them, giving an AC output proportional to displacement whose phase (0° or 180°) gives the direction. Phase-sensitive detection (see Fourier Transforms, Lock-in Detection, Box-car Integration and Modulation) recovers the signed output. LVDTs are frictionless, rugged and have essentially infinite resolution.

Capacitive sensors use \( C = \varepsilon A/d \): changing the gap gives \( \Delta C/C = -\Delta d/d \) (nonlinear in \( d \)), while changing the overlap area is linear. They reach sub-nanometre resolution.

Vacuum Gauges

No single gauge covers the whole vacuum range, so systems carry two or more. Orders of magnitude are what Part C tests:

GaugePrincipleApproximate working rangeGas dependence
Bourdon / diaphragmmechanical deformationatmosphere down to about 1 mbarnone
Capacitance manometerdiaphragm deflection read capacitivelyatmosphere to about \( 10^{-4} \) mbar (needs several heads)none: measures true pressure
McLeodcompresses a known gas volume, then reads with a liquid column (Boyle’s law)rough to high vacuum; absolute, used for calibrationcondensable vapours spoil it
Piraniheat loss from a heated wire by gas conduction, wire resistance read in a bridgeabout 10 to \( 10^{-3} \) mbaryes (thermal conductivity)
Thermocouple gaugesame heat-loss principle, wire temperature read by a thermocouplesimilar to Piraniyes
Penning (cold cathode)discharge in crossed electric and magnetic fields; current \( \propto \) pressurehigh vacuum, about \( 10^{-2} \) to \( 10^{-7} \) mbaryes
Hot-cathode ionisation (Bayard–Alpert)electrons ionise gas; ion current \( \propto \) pressurehigh to ultra-high vacuum, below about \( 10^{-3} \) mbar into \( 10^{-10} \) mbar and loweryes

Why the heat-loss gauges stop working at both ends: at high pressure the gas thermal conductivity becomes independent of pressure (mean free path shorter than the wire–wall gap), and at low pressure radiation and conduction through the supports swamp the gas contribution. The hot-cathode gauge’s low-pressure limit is set by soft X-rays from the electron collector releasing photoelectrons from the ion collector; the Bayard–Alpert design shrinks the collector to a fine wire to reduce this.

Magnetic Field

Hall probe

A current \( I \) through a slab of thickness \( t \) (along \( B \)) in a perpendicular field \( B \) develops a transverse voltage

$$\boxed{V_H = \frac{IB}{nqt} = R_H\frac{IB}{t}, \qquad R_H = \frac{1}{nq}}$$

The sign of \( V_H \) gives the carrier sign. Semiconductors (small \( n \)) give much larger Hall voltages than metals, which is why Hall probes use InSb, InAs, GaAs or Si. The response is linear in \( B \), works at DC, and needs calibration and temperature correction.

Other magnetic sensors

SensorPrincipleCharacter
Search (pick-up) coilFaraday induction, \( \mathcal{E} = -N\,d\Phi/dt \)AC fields only, or integrate the emf (fluxmeter)
Magnetoresistive (AMR, GMR)resistance depends on field direction or magnetisation alignmentsmall, sensitive; read heads
Fluxgatesaturation of a driven soft-magnetic core becomes asymmetric in an external field; second harmonic of the drive measures itsensitive DC vector magnetometer
SQUIDsuperconducting loop with Josephson junction(s); flux quantised in units of \( h/2e \)the most sensitive; needs cryogenics
NMR probeproton precession frequency \( \propto B \)absolute calibration standard

Vibration and Acceleration

A piezoelectric crystal (quartz, PZT ceramics) produces charge \( Q = d\,F \) proportional to applied force, with \( d \) the piezoelectric coefficient. Mounted against a seismic mass it becomes an accelerometer: \( F = ma \), so \( Q \propto a \). The charge leaks through the crystal and amplifier input resistance, so piezoelectric sensors cannot measure static forces; they are read with a charge amplifier (an op-amp integrator with feedback capacitor, \( V_o = -Q/C_f \)) that makes the output independent of cable capacitance.

A seismic (mass–spring–damper) sensor has resonance \( \omega_0 = \sqrt{k/m} \). Used below \( \omega_0 \) its relative displacement is proportional to acceleration (accelerometer); used above \( \omega_0 \) the mass stays nearly still in space and the relative displacement equals the base displacement (seismometer, vibrometer). Moving-coil velocity pickups output an emf proportional to velocity.

Optical Detectors

DetectorMechanismStrengthsLimits
Photodiode (PN, PIN, avalanche)internal photoeffect, \( I = R\,P \) with \( R = \eta q\lambda/hc \)linear, fast, cheapno gain (PN/PIN); APD adds gain with excess noise
Photomultiplier (PMT)external photoeffect at photocathode, then secondary-emission dynode chaingain \( \delta^n \sim 10^6 \)–\( 10^7 \); single-photon counting; fastfragile, high voltage, magnetic-field sensitive; cathode QE limits it to UV–visible–near IR
CCD / CMOS arraysphotoelectrons stored in pixel potential wells, then read outimaging, spectroscopy; cooled CCDs have very low dark currentreadout noise, finite well depth
Photoconductorconductivity rises with illuminationinfrared versions (narrow-gap semiconductors)slower; dark current needs cooling
Thermopilemany thermocouples in series sensing absorbed-heat temperature riseflat spectral response, absolute powerslow, low sensitivity
Bolometerresistance change of an absorbing elementbroadband IR to mm-wave; cryogenic versions very sensitiveslow (thermal time constant)
Pyroelectricchange of spontaneous polarisation with temperature gives a current \( \propto dT/dt \)room temperature, broadbandresponds only to changing light, so the beam is chopped
Golay cellgas expansion in a cell moves a mirrorfar IRslow, delicate

Photon detectors (photodiode, PMT, CCD, photoconductor) respond to photon number and have a long-wavelength cutoff \( \lambda_c = hc/E_g \) (or \( hc/\phi \)). Thermal detectors (thermopile, bolometer, pyroelectric, Golay) respond to absorbed power, have a flat spectral response, and are slower. Noise-equivalent power and detectivity \( D^* = \sqrt{A\,\Delta f}/\text{NEP} \) compare them; see the optoelectronics page.

The PMT gain: with \( n \) dynodes each multiplying by \( \delta \), \( G = \delta^n \); 10 dynodes at \( \delta = 4 \) give \( 4^{10} \approx 1.05 \times 10^6 \). Since \( \delta \) rises with the interdynode voltage, the gain is a steep function of supply voltage, so the high-voltage supply must be very stable.

Particle and Radiation Detectors

Gas-filled detectors

Radiation ionises the gas; an applied voltage collects the ions. The pulse size versus voltage defines the regions:

graph LR
    A[Recombination region: ions recombine] --> B[Ionisation chamber: all primary ions collected, no gain]
    B --> C[Proportional region: gas multiplication, pulse proportional to deposited energy]
    C --> D[Limited proportionality]
    D --> E[Geiger-Muller region: full avalanche, pulse size independent of energy]
    E --> F[Continuous discharge]
  • Ionisation chamber: current mode, used for dose and beam monitoring; no energy resolution from gain.
  • Proportional counter: gain from avalanche near a thin anode wire; pulse height still proportional to energy, so it gives energy information (soft X-rays) and can distinguish alpha from beta.
  • Geiger–Müller counter: each event triggers a discharge along the whole wire; large pulses, simple electronics, but no energy information and a long dead time while the discharge is quenched.

Dead-time correction

For a non-paralysable counter with dead time \( \tau \), true rate \( n \) and measured rate \( m \),

$$\boxed{n = \frac{m}{1 - m\tau}}$$

The two-source method measures sources 1, 2 and both together; for \( m\tau \ll 1 \), \( \tau \approx \dfrac{m_1 + m_2 - m_{12}}{2m_1m_2} \) (background neglected).

Scintillation detectors

Radiation excites a scintillator (inorganic crystals such as thallium-doped sodium iodide; organic plastics and liquids), which emits light read by a PMT or photodiode. Inorganic crystals have high density and atomic number, so they stop gamma rays efficiently and their light output is roughly proportional to energy. Organic scintillators are much faster but have low \( Z \), so they are used for timing, charged particles and fast neutrons.

Semiconductor detectors

A reverse-biased junction (silicon for charged particles and X-rays; high-purity germanium for gamma rays, cooled to reduce leakage current) acts as a solid ionisation chamber. Creating an electron–hole pair costs about 3.6 eV in silicon, against roughly 30 eV per ion pair in a gas and far more per collected photoelectron in a scintillator–PMT chain. More charge carriers per unit energy means smaller relative statistical fluctuation, so energy resolution is far better:

$$\frac{\sigma_E}{E} = \sqrt{\frac{F\,w}{E}}$$

where \( w \) is the energy per carrier pair and \( F \leq 1 \) the Fano factor (\( F = 1 \) gives pure Poisson statistics).

Memory Tricks

Don’t Confuse

ThisvsThat
RTD: positive, small, nearly linear coefficientNTC thermistor: negative, large, exponential
Thermocouple measures a temperature differenceRTD measures absolute temperature
Proportional counter: pulse height \( \propto \) energyGeiger counter: pulse height independent of energy
Photon detector: cutoff wavelength, fastThermal detector: flat response, slow
Pyroelectric: responds to \( dT/dt \), needs choppingThermopile: responds to steady power
Accelerometer: used below resonanceSeismometer: used above resonance
Hall probe: static fieldsSearch coil: changing flux only
Capacitance manometer: gas-independentPirani, ionisation gauges: calibrated per gas

Common Mistakes

Practice

All questions below are original practice questions written for this page. They are not from any CSIR NET paper.

Level 1: Foundation

  1. Name one active and one passive temperature transducer.
  2. Which vacuum gauge would you choose to read \( 10^{-9} \) mbar?
  3. A photomultiplier has 12 dynodes, each with secondary-emission ratio 4. What is its gain?
Answers
  1. Active: thermocouple. Passive: RTD or thermistor (also diode thermometer, which needs a bias current).
  2. A hot-cathode (Bayard–Alpert) ionisation gauge.
  3. \( 4^{12} \approx 1.7 \times 10^7 \).

Level 2: Part B Style

Practice Q1. A metal-foil strain gauge (gauge factor 2) is the only active arm of a Wheatstone bridge excited at 5 V. A strain of \( 500 \times 10^{-6} \) gives an output closest to:

(a) 0.63 mV (b) 1.25 mV (c) 2.5 mV (d) 5.0 mV

Answer: (b) 1.25 mV

\( \Delta R/R = G\varepsilon = 10^{-3} \); quarter bridge output \( = V_{ex}(\Delta R/R)/4 = 5 \times 10^{-3}/4 = 1.25 \) mV (the exact formula gives 1.249 mV). Option (c) is the half-bridge value, (d) the full bridge.

Practice Q2. A Hall probe of thickness 0.1 mm carries 10 mA in a field of 0.5 T. The carrier density is \( 10^{22} \) m\( ^{-3} \). The Hall voltage is about:

(a) 3.1 µV (b) 0.31 mV (c) 31 mV (d) 3.1 V

Answer: (c) 31 mV

\( V_H = IB/(nqt) = (0.01)(0.5)/(10^{22} \times 1.6 \times 10^{-19} \times 10^{-4}) = 0.005/0.16 \approx 0.031 \) V. A metal, with \( n \sim 10^{28} \) m\( ^{-3} \), would give a million times less, which is why Hall probes are semiconductors.

Practice Q3. A Geiger–Müller counter with a dead time of 100 µs records 5000 counts per second. The true count rate is:

(a) 2500 s\( ^{-1} \) (b) 5500 s\( ^{-1} \) (c) 7500 s\( ^{-1} \) (d) 10000 s\( ^{-1} \)

Answer: (d) 10 000 per second

\( m\tau = 5000 \times 10^{-4} = 0.5 \); \( n = 5000/(1 - 0.5) = 10\,000 \) s\( ^{-1} \). The counter is dead half the time. Option (a) applies the correction in the wrong direction.

Practice Q4. Which detector would be the poor choice for measuring a steady, unchopped infrared beam?

(a) thermopile (b) bolometer (c) pyroelectric detector (d) cooled photoconductor

Answer: (c) pyroelectric

The pyroelectric current is \( \propto dT/dt \); a steady beam produces a constant temperature after the initial transient and therefore no signal. It must be chopped, usually with lock-in detection at the chopping frequency.

Level 3: Part C Style

Practice Q5. An NTC thermistor reads 10.0 kΩ at 25.0 °C and 3.60 kΩ at 50.0 °C. Assuming the \( B \)-equation, consider:

  1. \( B \approx 3.9 \times 10^3 \) K.
  2. The temperature coefficient at 25 °C is about \( -4.4\% \) per kelvin.
  3. Measured with a 1 mA current, the thermistor at 25 °C dissipates 10 mW, which risks significant self-heating.
  4. The thermistor’s resistance is a linear function of temperature over this range.

Which are correct?

(a) 1 and 2 only (b) 1, 2 and 3 only (c) 2 and 4 only (d) 1 and 4 only

Answer: (b) 1, 2 and 3 only

Statement 1: \( B = \ln(R_1/R_2)/(1/T_1 - 1/T_2) = \ln(2.778)/(1/298.15 - 1/323.15) = 1.0217/2.595 \times 10^{-4} \approx 3937 \) K. True.

Statement 2: \( \alpha = -B/T^2 = -3937/298.15^2 \approx -0.044 \) K\( ^{-1} \). True.

Statement 3: \( P = I^2R = (10^{-3})^2 \times 10^4 = 10 \) mW. A small bead’s dissipation constant (power per kelvin of self-heating, quoted on its data sheet) is typically of the order of milliwatts per kelvin, so 10 mW would raise its temperature by kelvins, far more than the resolution sought. Low currents are used. True.

Statement 4: \( R \) is exponential in \( 1/T \). False.

Practice Q6. In a two-source measurement of a counter’s dead time, source 1 alone gives 8000 s\( ^{-1} \), source 2 alone 6000 s\( ^{-1} \), and both together 13 000 s\( ^{-1} \) (background negligible). The dead time is closest to:

(a) 1 µs (b) 10 µs (c) 50 µs (d) 100 µs

Answer: (b) 10 µs

\( \tau \approx (m_1 + m_2 - m_{12})/(2m_1m_2) = (8000 + 6000 - 13\,000)/(2 \times 8000 \times 6000) = 1000/(9.6 \times 10^7) \approx 1.04 \times 10^{-5} \) s.

Check: \( m\tau \) is up to 0.14, so the approximate formula is only rough (the exact two-source expression \( (m_1 + m_2 - m_{12})/(m_{12}^2 - m_1^2 - m_2^2) \) gives about 14 µs), but both point to option (b). The “lost” 1000 counts per second exist because the combined source is counted at a higher rate, where a larger fraction of events falls in the dead time.

Practice Q7. A gamma-ray line at 1 MeV is measured with (i) a gas proportional counter (\( w \approx 30 \) eV per ion pair) and (ii) a silicon detector (\( w = 3.6 \) eV per pair). Ignoring electronic noise and taking \( F = 1 \) for both, consider:

  1. The silicon detector produces about 8 times more charge carriers.
  2. Its statistical relative resolution is better by a factor of about \( \sqrt{8} \approx 2.9 \).
  3. The Poisson-limited FWHM for silicon is about 0.45%.
  4. Using a real Fano factor below 1 would make both resolutions worse.

Which are correct?

(a) 1 and 2 only (b) 1, 2 and 3 only (c) 3 and 4 only (d) 1, 2, 3 and 4

Answer: (b) 1, 2 and 3 only

Statement 1: \( N_{Si}/N_{gas} = 30/3.6 \approx 8.3 \). True.

Statement 2: relative resolution \( \propto 1/\sqrt N \), so the ratio is \( \sqrt{8.3} \approx 2.9 \). True.

Statement 3: \( N = 2.78 \times 10^5 \), \( 2.355/\sqrt N = 0.45\% \). True.

Statement 4: \( \sigma_E/E = \sqrt{Fw/E} \); \( F < 1 \) reduces the spread. False.

Quick Revision Table

SensorLawKey point
Pt RTD\( R = R_0(1 + \alpha T) \), \( \alpha \approx 0.00385 \) /°Cfour-wire connection
NTC thermistor\( R = R_0e^{B(1/T - 1/T_0)} \), \( \alpha = -B/T^2 \)very sensitive, nonlinear, self-heating
Thermocouple\( V \approx S(T_h - T_{\text{ref}}) \), tens of µV/Kneeds cold-junction compensation
Si diodeabout \( -2 \) mV/K at fixed currentcryogenic thermometry
Strain gauge\( \Delta R/R = G\varepsilon \); metal \( G \approx 2 \), semiconductor \( \sim 100 \)quarter/half/full bridge \( V_{ex}G\varepsilon\times(1/4, 1/2, 1) \)
LVDTAC output \( \propto \) displacement, phase gives signphase-sensitive detection
Hall probe\( V_H = IB/nqt \)semiconductors, DC fields
SQUIDflux quantum \( h/2e \)most sensitive magnetometer
Piezo accelerometer\( Q \propto F = ma \)no DC response; charge amplifier
PMT\( G = \delta^n \)single-photon counting
GM dead time\( n = m/(1 - m\tau) \)true rate > measured
Detector resolution\( \sigma_E/E = \sqrt{Fw/E} \)Si 3.6 eV, gas about 30 eV per pair
VacuumBourdon → Pirani → Penning → Bayard–Alperthigh to ultra-high vacuum needs ionisation

Key Takeaways

What’s Next