All lessons

Module 7: Electronics and experimental methods · Lesson 7.3 · about 55 min

Transducers and detectors

The sensors NET asks about: thermocouples, platinum resistance thermometers and thermistors, strain gauges and Wheatstone bridges, piezoelectric, Hall and LVDT sensors, first-order response, photodiodes, photoconductors and photomultipliers, thermal detectors, gas counters, scintillators and semiconductor detectors with dead time and the Fano factor, and vacuum gauges and pumps.

Key points, meanings and formulas on one page: choose "Save as PDF" in the print box.

Module 7 · about 55 minutes. By the end you will be able to read a thermocouple with cold-junction compensation, compare resistance thermometers, get a voltage out of a strain gauge with a bridge, work with piezoelectric, Hall and LVDT sensors, describe the speed of a first-order sensor, choose and use photodetectors, handle gas counters, scintillators and semiconductor detectors (including dead time and energy resolution), and pick the right vacuum gauge and pump. The semiconductor physics behind photodiodes is in the JAM lessons on semiconductors and diodes.

A transducer turns the quantity you care about (temperature, strain, light, radiation, pressure) into an electrical signal. For each one, NET wants three things: the physical principle, the size of the signal (so you know it needs a bridge or an amplifier), and the main way it goes wrong.

1. Temperature

Thermocouples (Seebeck effect). Two different metals joined at both ends give an EMF that depends on the difference of the junction temperatures, about 41 μV/°C for type K (chromel-alumel). Two laws do all the work:

  • Intermediate temperatures: E(T1,T3)=E(T1,T2)+E(T2,T3). Tables are made with the reference at 0 °C; if your reference sits at room temperature, add E(Troom,0) back (cold-junction compensation).
  • Intermediate metals: a third metal (copper leads, solder) adds nothing if both its ends are at the same temperature.

Worked example. Tables give E(300 °C)=12.209 mV and E(25 °C)=1.000 mV. With the reference at 25 °C the meter reads 12.209−1.000=11.209 mV. Using a constant sensitivity instead would give 0.041×275=11.28 mV, close but not exact, because the real curve is not straight.

Resistance thermometers. Platinum (Pt100: 100 Ω at 0 °C) is nearly linear, R=R0(1+αT) with α=0.00385 K⁻¹, stable and accurate. An NTC thermistor (a semiconductor) follows

R=R25exp⁡[B(1T−1T25)],1RdRdT=−BT2

about −4% per kelvin near room temperature, ten times platinum, but exponential and limited in range. Read resistance sensors with four wires (two carry current, two sense voltage) so the leads drop out, and keep the current small: self-heating is I2R divided by the dissipation constant.

Response time. Most sensors are first order: τdTs/dt=T−Ts. After a step they reach 63% of the change in one time constant and 99% in 4.6τ; following a steady ramp at rate a they lag by aτ.

Response curves of four transducers. Thermocouple: the EMF falls by E(T_ref) as the reference junction warms. Pt100 against thermistor: linear against steep and curved. Silicon photodiode: rising as λ/1240, then cut off at the gap. Strain-gauge quarter bridge: exact output against the linear estimate as the gauge factor grows.

2. Strain, force, displacement and field

Strain gauges. Stretching a wire lengthens it and narrows it: ΔR/R=(1+2ν)ε+Δρ/ρ. The gauge factor GF=(ΔR/R)/ε is about 2 for metal foil (1.6 from geometry with ν=0.3, the rest from resistivity) and 50 to 200 for semiconductor gauges (piezoresistance). Changes are tiny (0.1% at 500 microstrain), so gauges go in a Wheatstone bridge:

Vout≈n4VsGFε(n=1,2,4 active arms, strains arranged to add)

A half bridge (gauges on the top and bottom of a bent beam, in adjacent arms) doubles the signal, cancels temperature changes and removes the non-linearity of the quarter bridge, whose exact output Vsx/(4+2x) (with x=ΔR/R) falls short of linear by about x/2.

Worked example. One 120 Ω gauge (GF = 2) in a 10 V bridge under 10−3 strain: x=2×10−3, Vout=10×2×10−3/4=5 mV; the gauge resistance changes by only 0.24 Ω.

Piezoelectric sensors. A force F on quartz releases a charge q=dF (d=2.3 pC/N; PZT ceramics hundreds of pC/N). The voltage q/C depends on the cable capacitance too, so charge amplifiers are used. The charge leaks away with time constant RC: piezo sensors measure changing forces and vibrations, never static ones.

Hall sensors give VH=IB/(net): thin, lightly doped semiconductors (InSb, GaAs) give tens of mV per tesla where a metal gives nanovolts. LVDTs (a movable iron core between a primary and two opposed secondaries) give an AC output whose amplitude grows with displacement and whose phase flips by 180° as the core crosses the centre; a phase-sensitive detector (see lesson 7.4) turns this into a signed reading. Capacitive sensors use C=ϵ0A/d: sub-nanometre displacements with a bridge.

3. Light

Photodiodes collect the electron-hole pairs made by absorbed photons. The responsivity is

ℛ=ηehν=ηλ[nm]1239.8 A/W

rising with wavelength (more photons per watt) until the photon energy drops below the gap: silicon stops at 1.1 μm, germanium at 1.8 μm, InGaAs at 1.7 μm (the choice for 1550 nm). Reverse bias (photoconductive mode) widens the depletion layer: faster, linear, but with dark current. Zero bias (photovoltaic mode) has no dark current and is quietest for faint, slow signals; the open-circuit voltage grows only as (kBT/e)ln⁡(1+Iph/I0).

Photoconductors (CdS, PbS) have gain G=τlife/ttransit: the contacts replace every carrier that leaves, so one photon can pass many electrons. Photomultipliers add a photocathode (quantum efficiency 20 to 30% in the blue) and n dynodes, each multiplying by δ: gain δn, 410≈106, enough to count single photons. Avalanche photodiodes give gains of about 100 with extra noise. Thermal detectors (bolometers, thermopiles, pyroelectrics) measure the heat of absorbed light: flat response over wavelength but slow; photon detectors are fast but have a cut-off.

Worked example. A photodiode with η=0.8 at 800 nm: ℛ=0.8×800/1239.8=0.52 A/W, so 10 μW gives 5.2 μA.

4. Ionising radiation

Gas detectors. As the voltage rises, a gas-filled tube goes through the ionisation-chamber region (all primary ions collected, gain 1), the proportional region (gas multiplication near the thin anode, pulse ∝ deposited energy) and the Geiger-Müller region (every event triggers the same full discharge: counting but no energy information). In air about 34 eV makes one ion pair.

Dead time. After each count a GM tube is blind for τ (100 to 300 μs). For a non-paralysable counter

n=m1−mτ

with m the observed and n the true rate. A paralysable detector obeys m=ne−nτ and peaks at 1/(eτ): beyond that a stronger source reads lower.

Scintillators and semiconductors. NaI(Tl) gives about 38 photons per keV; after light collection and photocathode efficiency only a few thousand photoelectrons remain for a 662 keV gamma ray, limiting the resolution to about 4% (6 to 7% in practice). A germanium detector makes one electron-hole pair per 2.96 eV (silicon 3.6 eV), so N=E/w is in the hundreds of thousands, and the spread is reduced further by the Fano factor F≈0.1:

FWHM=2.355FEw

About 1.1 keV at 662 keV, thirty times better than NaI, at the price of cooling to 77 K (germanium’s small gap gives too much leakage current at room temperature).

gauge factor
fractional resistance change per unit strain: about 2 for metals, 100 for semiconductors
responsivity
amperes of photocurrent per watt of light; quantum efficiency divided by the photon energy in eV
dead time
the time after each count during which the detector cannot register another; the lost fraction is mτ
Fano factor
the variance of the number of charge carriers divided by its mean; below 1 because the carriers are not independent

5. Vacuum: making and measuring it

Range (mbar) Pumps Gauges
1000 to 1 rotary vane, scroll Bourdon, diaphragm, capacitance manometer
1 to 10−3 rotary, roots Pirani (heat loss through the gas)
10−3 to 10−8 diffusion, turbomolecular (need backing) cold-cathode (Penning), hot-cathode ionisation
below 10−8 ion, cryo, titanium sublimation, after baking Bayard-Alpert ionisation gauge

Diffusion and turbo pumps cannot exhaust to atmosphere: their jet or blades work only against a forevacuum. Pirani and ionisation gauges read differently for different gases; capacitance manometers do not. The mean free path in air is about 6.6×10−3 Pa m divided by the pressure: 1 m at 6.6×10−5 mbar.

Pump-down time. With a constant speed S and no outgassing, Vdp/dt=−Sp gives t=(V/S)ln⁡(p0/p): each decade of pressure costs (V/S)ln⁡10. A 50 L chamber on 5 L/s goes from 1000 to 0.1 mbar in 92 s; below about 10−6 mbar, outgassing from the walls takes over and the simple formula fails.

“A thermocouple measures the temperature of its tip.” It measures the difference between its two junctions. Forget the reference junction and every reading is off by the room temperature.

“A detector with more gain is always better.” Gain amplifies the noise that came with the signal too. Energy resolution is set by the smallest number in the chain (photoelectrons in a scintillator, electron-hole pairs in a semiconductor), not by the amplifier.

  • Thermocouple EMF ∝ junction temperature difference; law of intermediate temperatures; cold-junction compensation.
  • Pt100: R=100(1+0.00385T) Ω; thermistor: R∝eB/T, coefficient −B/T2; four-wire reading; self-heating I2R/δ.
  • First-order sensors: 63% in τ, ramp lag aτ.
  • Strain gauge GF ≈ 2 (metal); bridge output (n/4)VsGFε; piezo q=dF, no DC; Hall VH=IB/net; LVDT phase gives the sign.
  • Photodiode responsivity ηλ/1239.8 A/W with a band-gap cut-off; PMT gain δn; photoconductor gain τ/ttr.
  • Dead time n=m/(1−mτ); paralysable maximum 1/eτ; resolution FWHM 2.355FEw; Ge (2.96 eV, cooled) beats NaI.
  • Vacuum: rotary → diffusion or turbo → ion; Pirani to 10−3 mbar, ionisation gauges below; t=(V/S)ln⁡(p0/p).

Practise: exam questions on this lesson

NET Part C style, with full solutions. More on the CSIR NET practice page; the Counting statistics helper handles count-rate errors and background subtraction.

Next: lesson 7.4: every one of these signals comes with noise, and the lock-in amplifier is how weak ones are measured.

Check yourself

1A type K thermocouple (41 μV/°C) has its hot junction at 400 °C and its reference junction at 25 °C. What EMF does it give?
2A counter with 200 μs dead time (non-paralysable) records 500 counts/s. What is the true rate?
3Why does a silicon photodiode stop responding beyond about 1.1 μm?

Sources

  • E. O. Doebelin and D. N. Manik, Measurement Systems: Application and Design, 6th ed. (2011), chapters 3, 4 and 8
  • G. F. Knoll, Radiation Detection and Measurement, 4th ed. (2010), chapters 4 to 6, 8 and 11
  • J. Moore, C. Davis, M. Coplan and S. Greer, Building Scientific Apparatus, 4th ed. (2009), chapters 3 and 6
  • A. Roth, Vacuum Technology, 3rd ed. (1990), chapters 2, 5 and 8
  • CSIR-UGC NET Physical Sciences syllabus, Part C unit on experimental techniques (transducers, detectors, vacuum)

Your ticks stay in this browser. Save them as a file on the course home.

Updated 2026-11-22. Not affiliated with CSIR, UGC or the National Testing Agency. Check the official CSIR NET website for the current pattern. All questions are original.