EN

Thermocouples

Operation, types, construction, installation and calibration

The thermocouple is the most widely used temperature sensor in industry: it is cheap, rugged, can be as thin as a hair and measures from −270 °C to over 2000 °C. Its weak point is accuracy, which depends heavily on how it is selected, installed and wired. This guide covers what you need to get it right. For calculations, use the thermocouple calculator.

How it works: the Seebeck effect

When a conductor is subjected to a temperature gradient, its free electrons drift from the hot end to the cold end and a small voltage appears between its ends. That voltage depends on the material: if two different conductors A and B are joined at one end (the measuring junction) and heated, an electromotive force (EMF) appears between the free ends that depends on the temperature difference. Thomas Seebeck discovered it in 1821.

\[ E = \int_{t_{\text{ref}}}^{t} \big(S_A(t) - S_B(t)\big)\,dt \;\approx\; S\,(t - t_{\text{ref}}) \]

\(S\) is the Seebeck coefficient of the pair, from a few µV/°C to about 80 µV/°C. An important detail: the EMF is generated along the wires, wherever there is a temperature gradient, not at the junction. That is why the wires must be homogeneous throughout the hot zone: a contaminated or deformed section generates a different EMF and distorts the measurement.

The three practical laws

  1. Homogeneous circuit: a homogeneous conductor generates no net EMF, even if it passes through zones at different temperatures.
  2. Intermediate metals: any other metal (terminals, connectors, the instrument copper) can be inserted without affecting the measurement, as long as both its ends are at the same temperature.
  3. Intermediate temperatures: \(E(t_1, t_3) = E(t_1, t_2) + E(t_2, t_3)\). This is what makes cold junction compensation and tables referred to 0 °C possible.

Standard types

IEC 60584-1 defines eight types, each identified by a letter. The positive conductor is always named first. The EMF tables of all of them are in the calculator.

TypePositive / negativeContinuous useShort-term useµV/°C at 100 / 500 / 1000 °CIEC colour
KNickel-chromium (Chromel) / nickel-aluminium (Alumel)0 to 1100 °C−180 to 1300 °C41 / 43 / 39green
JIron / copper-nickel (constantan)20 to 700 °C−180 to 750 °C54 / 56 / 59black
TCopper / copper-nickel (constantan)−185 to 300 °C−250 to 400 °C47 / – / –brown
ENickel-chromium / copper-nickel0 to 800 °C−40 to 900 °C68 / 81 / –violet
NNicrosil / Nisil0 to 1150 °C−270 to 1300 °C30 / 38 / 39pink
RPlatinum-13 % rhodium / platinum0 to 1600 °C−50 to 1700 °C7.5 / 11 / 13orange
SPlatinum-10 % rhodium / platinum0 to 1550 °C−50 to 1750 °C7.3 / 9.9 / 11.5orange
BPlatinum-30 % rhodium / platinum-6 % rhodium100 to 1600 °C0 to 1800 °C0.9 / 5 / 9grey

The ranges are indicative: the real limit depends on the wire diameter, the sheath and the atmosphere (see below).

Type K

The most widely used, thanks to its wide range and price. It works well in oxidising or inert atmospheres, but has three weaknesses: between about 250 and 600 °C it suffers hysteresis due to short-range ordering in the positive leg, which can reach several degrees; in oxygen-poor atmospheres (inside closed sheaths without insulation) above 800 °C green rot appears, selectively oxidising the chromium; and above 1000 °C it drifts over time. The negative leg is magnetic, which helps to identify the polarity.

Type N

Developed in the 1980s to correct the shortcomings of type K. More chromium and silicon in the positive leg and more silicon and magnesium in the negative form a protective oxide layer and eliminate the magnetic hysteresis. It is more stable in the long term and withstands nuclear radiation better. With a Nicrobell-type sheath alloy it approaches the stability of platinum types at a fraction of their cost.

Type J

High output and cheap. It tolerates reducing atmospheres and vacuum, but iron oxidises quickly above about 550 °C and becomes brittle with moisture at low temperatures. At 770 °C iron changes its magnetic structure and the curve shows a step, which is why it is not used above about 750 °C.

Type T

The best for low temperatures and laboratory work: very accurate between −200 and 200 °C. The positive conductor is copper, which makes connections easier, but its high thermal conductivity can distort readings in short probes. Above 370 °C the copper oxidises.

Type E

It has the highest sensitivity (up to 81 µV/°C), useful when measuring small temperature differences. It works well at cryogenic temperatures and in oxidising or inert atmospheres. It is not magnetic.

Types R and S (platinum)

Stable and accurate up to 1600 °C. Type S was for decades the standard of the international temperature scale; type R gives slightly more output and is preferred in industry. Their enemy is contamination: metal vapours, silicon or phosphorus attack the pure platinum of the negative leg, so they are protected with high-purity alumina insulators and sheaths. At high temperature rhodium can diffuse into the negative leg and reduce the output.

Type B

For the highest temperatures, up to 1800 °C. Its EMF is almost zero below 50 °C, which has a curious advantage: it hardly needs cold junction compensation. It is not used below about 600 °C, where its signal is too small.

Non-standard thermocouples

How to choose a thermocouple

As a general rule: type K or N for almost everything up to 1200 °C (N if stability matters), T for low temperatures, J for older equipment or reducing atmospheres up to 700 °C, E when a lot of signal is needed, R or S up to 1600 °C and B above that.

Maximum temperature by wire diameter

The thinner the wire, the sooner it degrades. Recommended upper limits for thermocouples protected by a sheath (ASTM E230):

Type3.25 mm (8 AWG)1.63 mm (14 AWG)0.81 mm (20 AWG)0.51 mm (24 AWG)0.33 mm (28 AWG)0.25 mm (30 AWG)
J760 °C590 °C480 °C370 °C370 °C320 °C
K and N1260 °C1090 °C980 °C870 °C870 °C760 °C
E870 °C650 °C540 °C430 °C430 °C370 °C
T–370 °C260 °C200 °C200 °C150 °C
R and S–––1480 °C––
B–––1700 °C––

Service life also depends on temperature: as an order of magnitude, in base metal types every 50 °C above 500 °C halves the life, and doubling the wire diameter multiplies it by 2 or 3. Heating and cooling cycles shorten it even further.

Construction

Measuring junction type

JunctionDescriptionAdvantagesDrawbacks
ExposedThe wires protrude from the sheath, unprotected.Very fast response.Only for non-corrosive gases without pressure.
InsulatedJunction welded inside the sheath, electrically isolated from it.Protected and floating: immune to ground loops. The most widely used.Slower response.
GroundedJunction welded to the tip of the sheath.Response about 50 % faster than insulated, withstands high pressure.Connected to earth: may pick up noise and create ground loops.

Mineral insulation

This is the most common industrial construction: the wires run inside a seamless metal sheath filled with compacted magnesium oxide (MgO), which insulates them from each other and from the sheath. It is made in diameters from 0.25 to about 10 mm, can be bent and formed and withstands pressure, vibration and aggressive environments. The insulation resistance between conductors and sheath must exceed 100 MΩ at room temperature; if it drops, leakage paths appear that distort the reading. MgO is hygroscopic: a badly sealed end absorbs moisture and the insulation falls.

Wire insulation

InsulationOperating temperatureNotes
PVC−30 to 105 °CCheap, moisture resistant. Extension cables.
PTFE (Teflon)−75 to 250 °C (300 °C short-term)Chemically inert, good moisture resistance.
Varnished glass fibre−50 to 480 °CWithout varnish it reaches 540 °C, and some fibres 800 °C.
Ceramic / silica fibreup to 1000 °C or moreFor furnaces; poor abrasion resistance.
Ceramic beads (alumina, mullite)up to 1600–1800 °CFor bare wires inside ceramic sheaths. High-purity alumina with platinum.

Sheaths and thermowells

The sheath protects the thermocouple from the medium. Its material is chosen according to temperature and atmosphere; the values in the table are indicative for continuous use in air.

MaterialApprox. max. temp.OxidisingReducingSulphurousNotes
Stainless steel AISI 304 / 316800–870 °CGoodFairFairGeneral purpose. 316 resists corrosion better.
Stainless steel AISI 3101100 °CGoodFairFairHigh temperature in air.
Stainless steel AISI 4461100 °CVery goodGoodGoodGood resistance to sulphur; brittle when cold.
Inconel 6001150 °CGoodFairPoorThe most used for mineral insulation. Do not use with sulphur above about 550 °C.
Incoloy 8001100 °CGoodFairGoodResists carburisation and sulphur better than Inconel.
Hastelloy X1200 °CVery goodFairVery goodOxidising atmospheres containing sulphur.
Ni-Cr-Si alloy (Nicrobell)1250 °CVery goodVery goodFairDesigned for types K and N: same expansion as the wires, no contamination.
Mullite1600 °CGoodGood–Good thermal shock resistance. Contains silica: do not use with platinum at high temperature.
Recrystallised alumina1800 °CVery goodGood–Gas-tight. The right choice for platinum thermocouples.
Silicon carbide1500 °CFairGood–Very resistant to abrasion and thermal shock; used as an outer sheath.

A thermowell is a sheath machined from a single piece that screws or bolts to the process. It allows the probe to be replaced without opening the process and withstands pressure and fluid velocity, but it adds mass and slows the response. The bore must fit the probe closely; contact is improved with conductive paste or spring-loaded probes that press against the bottom.

Cold junction compensation

Since the thermocouple measures a temperature difference, the temperature of the reference junction (where the wires change to copper) must be known. There are four ways to do it:

Compensation only works if the thermocouple wires reach the instrument terminals. That is why extension or compensating cable is used and not copper.

Installation

A good measurement depends more on the installation than on the sensor. The goal is for the sensor to be at the temperature of the medium, not of its support.

Response time

It is defined as the time the sensor takes to reach 63.2 % of a step change in temperature (the time constant). It depends mainly on the sensor mass and the medium: in stirred water, a 0.5 mm mineral-insulated thermocouple responds in hundredths of a second, a 3 mm one in less than a second and a 6 to 8 mm one in several seconds; in still air the times multiply. A grounded junction responds roughly 50 % faster than an insulated one, and a thermowell can add minutes.

Average temperature with several thermocouples

Calibration, drift and faults

The tolerances in the standard apply to new thermocouples. With use, oxidation, contamination and metallurgical changes make the EMF drift, especially in the zone exposed to the temperature gradient. That is why calibrating a used thermocouple in the laboratory does not always help: when the immersion depth changes, the degraded zone is somewhere else and the error changes. In practice, base metal types are replaced periodically and critical ones are checked in situ with a reference probe next to the sensor.

Common symptoms and causes

SymptomProbable cause
Over-range reading or "broken probe"Broken wire, loose connection or open measuring junction.
Reads ambient temperatureShort circuit in the cable or probe: the new "junction" is at the short.
Temperature falls when the process risesReversed polarity somewhere in the cable.
Constant error of several degreesThermocouple or cable type different from the one configured, joint with copper, or compensating cable at excessive temperature.
Unstable or jumpy readingElectrical noise, grounded junction with a ground loop, or moisture in the mineral insulation.
Slow drift over timeThermocouple degradation from temperature or contamination. Time to replace it.

An abnormally high loop resistance indicates wires thinned at the end of their life; a very low one, a short circuit. The calculator gives the expected resistance for the cable.

The International Temperature Scale ITS-90

Since 1 January 1990 temperature has been defined by the ITS-90 scale, which replaced IPTS-68. It sets a series of reproducible fixed points and the instruments for interpolating between them: gas and helium vapour pressure thermometers at very low temperatures, platinum resistance thermometers between 13.8 K and 961.78 °C, and Planck's radiation law above that. The type S thermocouple, which had been a defining instrument, became a secondary standard. All current thermocouple and Pt100 tables are referred to this scale.

Fixed pointT₉₀ (K)t₉₀ (°C)
Triple point of hydrogen13.8033−259.3467
Triple point of neon24.5561−248.5939
Triple point of oxygen54.3584−218.7916
Triple point of argon83.8058−189.3442
Triple point of mercury234.3156−38.8344
Triple point of water273.160.01
Melting point of gallium302.914629.7646
Freezing point of indium429.7485156.5985
Freezing point of tin505.078231.928
Freezing point of zinc692.677419.527
Freezing point of aluminium933.473660.323
Freezing point of silver1234.93961.78
Freezing point of gold1337.331064.18
Freezing point of copper1357.771084.62

Related tools: thermocouple calculator and tables, cable colours, Pt100 calculator and 4–20 mA transmitters.