Thermocouple

What is a Thermocouple

A thermocouple is a temperature sensor built from two wires of different metals or alloys, joined at one end to form the measuring junction (often called the hot junction). When that junction sits at a different temperature from the open ends (the reference or cold junction), a voltage appears across the open ends. This is the Seebeck effect, named after Thomas Johann Seebeck, who observed in the 1820s that a temperature difference across a loop of two dissimilar conductors produces a measurable electromotive force. Temperature, alongside pressure, flow, and level, is one of the four fundamental process variables, and the thermocouple is its most common field sensor.

The voltage is not generated at the junction itself but distributed along each wire wherever a temperature gradient exists, in proportion to the Seebeck coefficient of that material. The net signal at the terminals is the difference between the two legs integrated over the gradient, which is why a thermocouple inherently measures a temperature difference, not an absolute value. For type K this scale factor is roughly 41 microvolts per degree Celsius and is nearly constant from 0 to 1000 degrees Celsius. Type E is the most sensitive of the common types at about 61 microvolts per degree, followed by type J near 50 and type T near 39.

Because the published voltage-to-temperature tables in IEC 60584-1 and ASTM E230 are referenced to a cold junction held at 0 degrees Celsius (the ice point), every real instrument must perform cold junction compensation. It measures the temperature of the terminal block with a separate sensor, usually a thermistor or RTD, converts that temperature to its equivalent thermocouple voltage, and adds it to the measured signal. Without compensation, a junction truly at 100 degrees Celsius read on a terminal block at 25 degrees would report about 75 degrees. Modern transmitters and data loggers do this automatically, but the principle explains why thermocouple wiring discipline matters so much.

The practical appeal of the thermocouple is its combination of range, ruggedness, speed, and cost. The eight standardized types together span approximately -270 to +1820 degrees Celsius, far beyond any resistance thermometer. A thermocouple needs no excitation current, contains no fragile resistance element, survives vibration and thermal shock, and can be welded into a junction smaller than a grain of rice for fast response. The trade-offs are lower absolute accuracy than an RTD at moderate temperatures, a nonlinear and relatively weak signal, vulnerability to electrical noise, and gradual drift caused by metallurgical changes in the wires over their service life.

Four engineering metrics govern thermocouple quality in service: tolerance class (the calibration error band per IEC 60584-2), drift over time at temperature, the usable temperature range for the chosen type and atmosphere, and the junction and sheath construction that determines response time and mechanical protection. As with any sensor, the cheapest device rarely owns the lowest total cost: an under-specified type that drifts negative through green rot or fails by sheath oxidation forces unplanned shutdowns whose cost dwarfs the original wire savings.

Standardized Thermocouple Types

IEC 60584-1 and ASTM E230 standardize eight letter-designated types, split into base-metal types (J, K, T, E, N) and noble-metal types (R, S, B). Each type fixes a specific pair of alloys, and therefore a specific voltage curve, usable range, sensitivity, and preferred atmosphere. Selecting the type is the first and most consequential decision: it sets the achievable temperature ceiling, the drift behavior, and roughly the cost. The table below summarizes the standardized types with their alloy pairs and continuous-use ranges.

Type K (nickel-chromium versus nickel-aluminium) is the workhorse of industry and the default for general-purpose measurement to about 1260 degrees Celsius in oxidizing atmospheres. Chromel is roughly 90 percent nickel and 10 percent chromium; alumel is roughly 95 percent nickel with 2 percent aluminium, 2 percent manganese, and 1 percent silicon. Type K is inexpensive and widely stocked, but it suffers a known hysteresis and reversible short-range ordering effect in the 300 to 550 degree band and is vulnerable to green rot in low-oxygen atmospheres between roughly 815 and 1040 degrees.

Type J (iron versus constantan) is common on older plants and in reducing atmospheres or vacuum, where the iron leg is less harmful than in type K. Its hard ceiling is 760 degrees Celsius: above that the iron passes through a magnetic transformation that permanently de-calibrates the wire, so type J must never be pushed higher. Type T (copper versus constantan) is the precision base-metal choice for cryogenic, laboratory, and food duty from -200 to 350 degrees, with the tightest base-metal tolerance and good behavior in moist environments because copper resists corrosion.

Type N (nicrosil versus nisil) was developed specifically to overcome type K weaknesses. The higher silicon and chromium content forms a stable protective oxide that resists green rot and the ordering drift, so type N holds calibration far better than type K above 500 degrees and is increasingly specified for new high-temperature installations to about 1260 degrees. Type E (chromel versus constantan) offers the highest sensitivity of the standard types, which suits low-temperature and low-signal applications, and is non-magnetic.

The noble-metal types R, S, and B use platinum and platinum-rhodium alloys for the highest temperatures. Types R and S serve continuously to about 1600 degrees Celsius and briefly higher in clean oxidizing or inert atmospheres, and type S is the historical reference for the international temperature scale. Type B uses Pt-30%Rh against Pt-6%Rh and reaches about 1700 degrees, with the unusual property that below roughly 50 degrees its output is negligible, so it needs no cold junction compensation in normal use. All three are expensive, low-output, and intolerant of reducing atmospheres, where rhodium migration embrittles and destroys the wire.

Junction and Sheath Construction

Beyond the alloy type, the physical construction of the probe sets its response time, mechanical strength, and electrical behavior. The dominant industrial form is the mineral-insulated metal-sheathed (MIMS) thermocouple, standardized by IEC 61515. The thermoelement wires run inside a metal sheath, electrically separated from each other and from the sheath by tightly compacted magnesium oxide (MgO) powder. The assembly is swaged or drawn down to compress the MgO into a dense ceramic, which gives flexibility, pressure containment, and high insulation resistance. A common sheath material is Inconel 600, a nickel-chromium-iron alloy that resists oxidation to high temperature.

Three junction configurations are available, and the choice trades response speed against electrical isolation. The table below compares them; the differences directly affect both measurement accuracy in noisy plants and the risk of damaging ground loops.

In a grounded junction the welded bead is bonded to the inside of the sheath tip. Heat transfers directly through the sheath wall, so response is fast and the probe withstands pressure, but the measurement circuit is now electrically common with the sheath. If the process metal is earthed at one point and the instrument at another, the two ground potentials can differ and drive a circulating current through the loop, producing reading errors or, in extreme cases, equipment damage. Grounded junctions are the usual choice for fast control loops where the sheath can be kept at a single ground reference.

An ungrounded junction keeps the welded bead floating inside the sheath, separated by the MgO insulation. Response is slower because heat must cross the insulation gap, but the floating measurement circuit rejects common-mode noise and eliminates ground-loop risk, which is why most general-purpose industrial probes ship ungrounded. An exposed junction leaves the bare bead protruding from the sheath end for the fastest possible response, but it is open to corrosion, condensation, and mechanical damage, so it is restricted to clean, dry, non-pressurized gas measurement such as air ducts.

Probe response time scales strongly with sheath diameter and junction style. A thin exposed bead can respond in well under a second, while a large ungrounded sheathed probe inside a thermowell may take tens of seconds to settle. For comparison, a grounded thermocouple junction typically responds several times faster than an equivalent Pt100 RTD, because the RTD cannot be grounded and always carries its insulation thermal lag. Where the process is corrosive, abrasive, or high pressure, the probe is installed inside a thermowell, which protects the sheath and allows the sensor to be removed without breaking process containment, at the cost of added thermal lag.

Wiring construction matters as much as the probe. The loop from sensor to instrument must be continued in matching thermoelement alloys, using extension grade wire for base-metal types or compensating cable for the costly noble-metal types R, S, and B. Any junction to ordinary copper before the cold junction compensation point becomes a parasitic thermocouple and injects error. Polarity and standard discipline are covered in the materials chapter, but the construction rule is simple: keep the same metals all the way to the compensated terminal block.

Materials, Atmospheres and Standards

A thermocouple is only as good as its compatibility with the process atmosphere. The two most common silent failure modes are green rot in type K and rhodium embrittlement in the noble-metal types, both driven by the wrong atmosphere rather than overtemperature. Matching the alloy and sheath to the atmosphere is therefore a first-order selection decision.

Green rot is preferential oxidation of the chromium in the chromel (positive) leg of a type K thermocouple when oxygen is present but scarce, typically inside an imperfectly sealed sheath between about 815 and 1040 degrees Celsius. As chromium depletes, the leg loses output and the reading drifts negative, sometimes by tens of degrees, with no outward sign. The cure is to remove the marginal oxygen (fully oxidizing or fully inert atmosphere), use a larger wire gauge, vent trapped moisture, or switch to type N, whose nicrosil and nisil alloys form a self-protecting silicon-rich oxide layer that resists the mechanism.

Reducing atmospheres are the principal enemy of platinum types R, S, and B. Hydrogen, carbon, or metal vapors attack the platinum lattice, and rhodium can volatilize from the positive leg and migrate to the pure platinum leg, forming a brittle alloy that snaps. For this reason noble-metal types are specified for clean oxidizing or inert service only and are sheathed in dense alumina or platinum-rhodium tubes rather than left bare. Type J, by contrast, tolerates reducing atmospheres and vacuum better than type K because it has no chromium to oxidize, which is one reason it persists on furnace and vacuum applications below 760 degrees.

Color codes are a frequent source of wiring error because two incompatible systems coexist worldwide. The IEC 60584-3 system and the North American ANSI MC96.1 system assign different colors and even disagree on which leg is negative, so the same green jacket means type K under IEC but does not exist under ANSI. The table below lists both systems for the base-metal types; verify which standard a cable follows before terminating it.

Under IEC 60584-3 the negative leg is always white and the positive leg and outer jacket carry the type color. Under ANSI MC96.1 the negative leg is always red, the positive leg carries the type color, and thermocouple-grade jackets are brown while extension-grade jackets match the type. Reversing polarity does not merely flip the sign of the small junction error; it doubles the cold-junction error, so confirming polarity against the correct standard is essential.

The governing standards form a small, coherent family. IEC 60584-1 defines the reference voltage functions; IEC 60584-2 (now folded into the -1 edition) defines tolerance classes 1, 2, and 3; IEC 60584-3 covers extension and compensating cable tolerances and color codes; and IEC 61515 governs mineral-insulated metal-sheathed cable and probes. In North America, ASTM E230 publishes the equivalent EMF tables and ANSI MC96.1 the limits of error and color codes. A type K wire that is compliant with these references reads identically on any compliant instrument, which is the entire value of standardization.

Key Specification Parameters

Thermocouple datasheets are short compared with transmitters, but a handful of parameters drive the selection. The most important is the tolerance class, because it sets the calibration error band the maker guarantees. IEC 60584-2 defines tolerance as the greater of a fixed value in degrees Celsius and a fraction of the absolute measured temperature, so the permissible error grows with temperature. The table below gives the exact class 1 and class 2 limits for the common base-metal types.

Tolerance class is read as the larger of the two expressions. For a type K class 1 element at 1000 degrees Celsius, the limit is 0.004 times 1000, or plus-or-minus 4 degrees; for class 2 it is 0.0075 times 1000, or plus-or-minus 7.5 degrees. Type T is the tightest base-metal type, reaching plus-or-minus 0.5 degrees in class 1 near ambient. There is also a class 3 intended for cryogenic use down to -200 degrees, with wider bands at the cold extreme. Class 1 typically costs roughly twice class 2, so it should be specified only where the overall loop accuracy budget genuinely needs it.

Sensitivity (the Seebeck coefficient) determines how much signal the instrument receives per degree and therefore the achievable resolution and noise immunity. Type E leads at about 61 microvolts per degree, type J about 50, type K about 41, and type T about 39, while the noble-metal types R, S, and B output only about 10 microvolts per degree, which is one reason they demand high-quality, low-noise input electronics. Higher sensitivity also relaxes the demands on the cold junction compensation circuit.

Drift and long-term stability describe how the calibration shifts over service life from metallurgical change, oxidation, and contamination. Type K drifts noticeably above 500 degrees from ordering effects and green rot; type N was engineered to suppress both and holds calibration far better at the same temperatures; noble-metal types are the most stable in clean oxidizing service and are used as reference standards. Drift is a service-life specification, not a fresh-out-of-box one, and is the main reason high-temperature thermocouples are recalibrated or replaced on a schedule.

Response time is set mainly by junction style and sheath diameter, as covered in chapter 3: exposed beads respond in well under a second, while large ungrounded probes in thermowells take tens of seconds. Temperature range must be read together with the atmosphere, because the safe ceiling drops sharply outside the preferred atmosphere. Finally, insulation resistance of the MgO at temperature matters for MIMS probes: at high temperature the insulation resistance falls, which can shunt the signal and cause errors if the probe is very long or run hot for extended periods.

Selection Decision Factors

To convert the preceding chapters into a specific part number, work through the decision sequence below in order. Most thermocouple failures trace not to a single wrong parameter but to skipping the atmosphere check or mismatching the extension cable, so treat each step as a gate before moving on. These steps can serve as a fixed RFQ template.

1. Temperature range and atmosphere: Establish the maximum continuous and peak temperatures, then the atmosphere (oxidizing, reducing, inert, or vacuum). The pair selects the type: K or N for high-temperature oxidizing, J for reducing or vacuum below 760 degrees, T for cryogenic and food, and R, S, or B for clean oxidizing service above 1300 degrees.
2. Tolerance class: Choose class 1 only where the loop accuracy budget needs it, otherwise class 2. Read the class limit at your actual operating temperature using the IEC 60584-2 formula, not just the headline degrees value.
3. Junction configuration: Grounded for fast response with single-point grounding, ungrounded for noise immunity and ground-loop avoidance, exposed only for clean dry low-pressure gas. Most industrial probes are ungrounded by default.
4. Sheath material and diameter: Match the sheath alloy (for example Inconel 600) to the process chemistry and temperature, and size the diameter against the required response time and mechanical strength. Smaller diameter responds faster but is more fragile.
5. Thermowell and process connection: Decide whether a thermowell is required for pressure, corrosion, abrasion, or hot-swap maintenance, then specify the connection (threaded, flanged, or sanitary clamp) and insertion length so the junction reaches the representative process zone.
6. Extension wiring and color standard: Specify extension grade wire for base-metal types or compensating cable for R, S, B, all in the correct IEC 60584-3 or ANSI MC96.1 colors, and confirm polarity. Never substitute plain copper before the cold-junction point.
7. Cold junction compensation and output: Confirm the receiving instrument performs cold junction compensation and accepts the chosen type, or specify a head-mounted transmitter that converts to 4-20 mA, HART, or a digital bus close to the sensor to avoid running long thermocouple wire through noisy areas.
8. Certifications and total cost of ownership: Add any required hazardous-area approvals (ATEX, IECEx, NEPSI), functional safety (SIL), and sanitary or aerospace specifications, then weigh purchase price against recalibration interval, drift-driven replacement, and downtime risk. A type N element that resists green rot can outlast several type K replacements in the same furnace.

One last dimension is manufacturer serviceability: availability of matched extension cable, calibration certificates traceable to a national standard, stock of spare elements and thermowells, and local recalibration service. WIKA, Endress+Hauser, Emerson, Yokogawa, Honeywell, ABB, and specialist makers such as DwyerOmega and Watlow supply standardized thermocouples with calibration documentation and field service networks, which matters most for high-temperature furnace and reactor duty where elements are consumable and downtime is costly. Confirm that the supplier can provide the specific type, tolerance class, and certification your loop requires before committing.