Product type
A thermocouple measures temperature by generating its own voltage. Two dissimilar metals joined at one point produce a small potential difference when that junction sits at a different temperature from the other end of the wires, and the size of that voltage follows a published curve for the pair of metals used. Nothing has to be supplied to the sensor for it to work, which is the property that decides where thermocouples are used: at the top of the temperature range, at the smallest measuring points, and on machines where a resistance element would not survive.
A thermocouple measures temperature by generating its own voltage. Two dissimilar metals joined at one point produce a small potential difference when that junction sits at a different temperature from the other end of the wires, and the size of that voltage follows a published curve for the pair of metals used. Nothing has to be supplied to the sensor for it to work, which is the property that decides where thermocouples are used: at the top of the temperature range, at the smallest measuring points, and on machines where a resistance element would not survive.
The letter types are the shorthand for those metal pairs. Type K is the general-purpose workhorse across the moderate to high range, type J serves older equipment, type T covers low temperatures and refrigeration, type E offers the highest sensitivity of the base-metal pairs, type N was developed for better stability at temperature, and types R, S and B use platinum and rhodium for the highest ranges and the most demanding duty. The choice between them is a choice of range, sensitivity, atmosphere and cost, and it fixes what the receiving instrument has to be set to.
Buyers reach for a thermocouple in two situations that look different but are the same problem. The first is a temperature beyond what a resistance element survives, typically above 500°C and certainly above 850°C. The second is a measuring point too small, too fast or too awkward for an inserted probe: an injection moulding barrel, a heating band, a bearing housing, an exhaust duct, a glass or ceramic surface. In both, the deciding factor is usually mechanics rather than the element.
What separates one thermocouple assembly from another is rarely the letter. It is the junction form, the sheath material and diameter, how the probe is held against the process, and how the cold junction at the far end is handled. Those four decisions set the response time, the life and the actual measurement error, and they are the ones worth settling before price is discussed.
The effect is named after Seebeck. Where two dissimilar metals meet, a temperature difference between that junction and the other ends of the wires drives a small current, and the resulting voltage is roughly proportional to the temperature difference over a limited span. It is not proportional over the whole range, which is why each letter type has its own polynomial and why the instrument has to be told which type it is reading. A thermocouple signal is also small: tens of microvolts per degree for the common base-metal types, and only a few microvolts per degree for the platinum types.
Because the voltage is generated between the measuring junction and the terminals where the wires land, the temperature at those terminals is part of the measurement. The remedy is cold junction compensation: the instrument measures the temperature at its own terminals and adds the corresponding voltage back, so that the reading reflects only the measuring junction. Compensation is not optional and it is not a refinement. Without it a thermocouple reports the difference between two temperatures, and only one of them is the one wanted.
The same principle constrains the wiring. Every place where thermocouple wire meets copper is itself a thermocouple junction, so the circuit has to be made up in the matching thermocouple wire and, where it is extended, with compensating or extension cable of the same type and with connectors made of the same alloys. A copper terminal block between the probe and the instrument does not break the measurement so much as insert two more junctions whose temperatures are unknown, and the error they introduce is fixed and easy to mistake for sensor drift.
Two consequences follow for the junction itself. A grounded junction welds the measuring point to the sheath, which gives fast response because heat and current both pass easily, but also connects the measuring circuit to whatever the sheath touches; on earthed metal that path can be a source of noise or of circulating current. An ungrounded junction insulates the measuring point inside the sheath, which isolates the circuit at the cost of response time. An exposed junction is faster still and is used where the medium is not corrosive and does not need to be kept out.
Most of the specification is fixed by these decisions rather than by the accuracy class. Each one constrains the next, which is why they are settled in order.
| Decision | Options to fix | What it actually decides |
|---|---|---|
| Calibration type | K, J, T, E, N, R, S or B, and the corresponding extension cable | The usable range, the sensitivity in microvolts per degree, the atmosphere the pair tolerates and what the receiving instrument must be set to. It is the one decision that invalidates the reading outright if it is wrong. |
| Junction form | Grounded, ungrounded or exposed; and the junction may be welded or part of a spring-loaded contact | The response time, the electrical isolation between the measuring circuit and the machine, and how much protection the junction has from the medium. |
| Sheath and diameter | Stainless steel, Inconel, ceramic or a braided mineral-insulated cable, in diameters from under a millimetre to several millimetres | How fast the assembly responds and what it survives. A heavy sheath in a corrosion-resistant alloy lasts, and it responds like the sheath rather than like the junction inside it. |
| How the probe is held | Spring clamp, threaded head, compression fitting, weld pad or bare tip against a surface | The contact pressure and the immersion depth, which together decide whether the junction sees the process temperature or the temperature of the bracket holding it. |
| Cold junction handling | A terminal head at the probe, a transmitter mounted at the probe, or compensating cable run back to the instrument | Where compensation happens and what the run costs. Transmitting at the probe converts the small voltage to a loop signal before it can pick up interference. |
| Number of measuring points | One probe per channel, or a multiplexed input card | Whether every point needs its own instrument channel and its own compensation, which on a multi-point machine is often the larger part of the cost. |
| Factor | How to settle it | Why it decides the answer |
|---|---|---|
| The top of the temperature range | Take the normal working temperature and add the overshoot the process actually produces, not the overshoot the control loop is supposed to produce. | The letter type and the sheath material both follow from it. A base-metal pair chosen for economy stops being the right answer well below the range printed for it once the atmosphere is taken into account. |
| The atmosphere | State what the probe is in: air, flue gas, molten metal, a vacuum, a reducing or oxidising environment, or a hot surface in still air. | Type K is prone to drift in a reducing atmosphere and to the short-range ordering effect around 250 to 550°C; type N was developed in part to answer that. Platinum types are the answer at high temperature but are not immune to contamination. |
| Response time required | Decide whether the reading has to follow a changing process or is only ever a steady-state value. | A fine sheath, a grounded or exposed junction and a small mass at the tip all shorten the response, and each trades protection or isolation for speed. A slow assembly reading a cycling process reports an average that nobody asked for. |
| Electrical environment | Say whether the probe is fixed to earthed metal, and whether variable-speed drives or heaters are nearby. | The microvolt signal picks up interference that a 4-20 mA loop shrugs off. Screened or braided lead and an ungrounded junction are the usual answers, and transmitting at the probe is the more complete one. |
| Distance to the instrument | Measure the run, then decide where the cold junction should be. | Compensating cable is more expensive than copper and has its own tolerance, so a long run is usually better served by a transmitter at the probe than by a long compensating lead back to a control room. |
| Whether the reading must be verifiable | Ask whether the value will be printed, audited or used to release a batch. | A bare thermocouple carries no traceable calibration unless one is provided with it. Where the number has to be defended, the assembly needs a documented tolerance and a route to re-verification. |
These are the types an enquiry for a thermocouple usually turns out to be, and the point at which the choice is genuinely decided.
| Nearest alternative | What separates it from a thermocouple |
|---|---|
| Type K thermocouples | A single letter type rather than the family. Worth choosing where the general-purpose pair covers the range and the atmosphere, and worth avoiding where the process sits in the ordering range or runs reducing. |
| RTD temperature sensors | Measures resistance rather than generating a voltage, so it needs an excitation current and gives a slower, larger and more consistent signal. Where the temperature is within the range of a platinum element and the reading has to agree between instruments, the RTD is the better instrument and the thermocouple is not an equivalent. |
| Resistance temperature detectors | The bare resistance element or sensing component, supplied as a leaded part rather than as a fitted assembly with a process connection. Chosen where the housing and connection are the buyer’s own. |
| Temperature transmitters | Measures nothing by itself. It reads the thermocouple and converts the small voltage into a loop signal before it travels, which is how a thermocouple is normally connected when the instrument is more than a few metres away. |
| Infrared thermometers | Reads a surface without contact, so it can measure a moving or inaccessible target that no probe could touch. It measures a surface rather than the bulk, and emissivity has to be known, so it complements a thermocouple rather than replacing it. |
| Temperature controllers | Takes the thermocouple signal and drives a heater or cooler towards a set point. It is the instrument the probe is connected to on a machine, and it is the part that has to be set to the same letter type. |
The letter type is what an enquiry names first and what decides least. Junction form, sheath and how the probe is held decide whether the measurement works.
The first reply can be a quotation rather than a set of questions if the enquiry carries these. Where a figure is not known, send the duty description and we will confirm what it implies.
Available products
Thermocouples
Spring-clamp thermocouple with a stainless steel M12×1.5 threaded head and a braided high-temperature shielded lead, for moulds, extruders, heating bands and other points that need a probe pressed into place.
Also in Temperature Measurement
Every type we quote for under Temperature Measurement. A type with no link yet is still quoted for - it is described on the Temperature Measurement overview, and gets a page of its own as soon as a model is published against it.
Frequently asked
Type K covers the widest useful span with the best availability and is the sensible default below about 1100°C in clean oxidising conditions. Move away from it for a reason rather than by preference: type T for low and sub-zero temperatures, type J where existing equipment is wired for it, type N where stability in the moderate range matters, type E where a stronger signal helps, and a platinum type where the temperature or the environment rules out a base-metal pair.
Because the junction box is part of the circuit. Where thermocouple wire meets copper, a further thermocouple junction is created, and if that junction sits at a different temperature from the instrument terminals the compensation no longer matches the real circuit. The usual causes are a copper terminal block in the run, the wrong extension cable, or terminals made of brass rather than the matching alloy.
Only if the extension is treated as part of the measurement. Copper cable moves the cold junction to the far end of the copper, and the instrument then compensates for the wrong point. Extension or compensating cable of the matching type is the correct answer, and connectors made of the same alloys. On a long run a transmitter at the probe usually costs less than the compensating cable and removes the problem entirely.
It is stated as a tolerance class on the wire itself and it applies at a reference temperature, not across the whole range, so a single accuracy figure for a thermocouple assembly is misleading. On top of that sit the cold junction compensation error in the instrument, the drift of the wire in service and, where the probe is not immersed, the difference between the tip and the medium. Where the reading has to be defensible, the whole assembly is what has to be toleranced and verified.
Send the specification and we will confirm the matching model, availability and lead time. Minimum order quantity is 1 pc.
Request for quotation
Send the technical detail and we will confirm availability, lead time and terms. In-stock items ship within 48 hours; customized specifications typically take 2-10 days. Minimum order quantity is 1 pc.