Thermocouple Extension Wire and Compensating Cable

By Wu Peng, Senior Process Instrumentation Engineer · Last reviewed September 18, 2026

Thermocouple extension wire carries the signal from the sensor head back to the instrument using the same alloy pair as the thermocouple itself, so the circuit never meets a new pair of metals until it reaches the instrument terminals. It is designated with an X after the type letter: KX, JX, TX, EX, NX. Compensating cable does the same job with cheaper substitute alloys, and is only valid while the joint at the head stays below a stated temperature.

The reason this matters is arithmetic rather than principle. Splice ordinary copper into a Type K circuit at a head sitting at 100 °C and the reading falls by about 76 °C. The splice, not the instrument, has become the reference junction.

This page covers which cable to specify and what the grades guarantee. It also covers how far a given gauge will run before the instrument objects, and what a reversed pair looks like on the display.

Contents

• Extension or compensating cable
• Grades and error limits
• Why copper fails
• Resistance by wire gauge
• Run length by gauge
• Insulation and jacket
• Polarity and wiring errors
• Ordering and installation
• FAQ

Extension or compensating cable

Extension cable uses the same alloys as the thermocouple. Compensating cable uses a different, cheaper alloy pair chosen to produce nearly the same voltage as the thermocouple over a limited temperature band. The distinction only matters at one place: the joint where the cable meets the sensor leads, usually inside the head.

Keep that joint inside the band and compensating cable behaves like extension cable at a fraction of the cost. Let the joint run hot and the substitute alloys stop tracking. The error then appears in the reading, with nothing on the display to suggest a cable problem.

Type	Extension code	Compensating code	Joint temperature limit
K	KX	KCA, KCB	KCB below 100 °C
N	NX	NC	Below 150 °C
J	JX	None offered	Extension grade only
T	TX	None offered	Extension grade only
E	EX	None offered	Extension grade only
R	Not available	RCA, RCB	RCA below 100 °C; RCB below 200 °C
S	Not available	SCA, SCB	SCA below 100 °C; SCB below 200 °C
B	Not available	Copper pair	Copper usable below 100 °C

SCA is the same material as RCA and SCB the same as RCB, which is why one drum often serves both types. Above 100 °C a Type B joint needs a matched alloy rather than plain copper.

Two rows in that table are worth a second look. Catalogues will sell you something listed as type S thermocouple extension wire, but what ships is compensating cable, because there is no true extension cable for Type R or Type S. Platinum alloy cable would cost more than the sensor, so compensating cable is the normal method for both.

The naming makes it worse. American catalogues list the R and S compensating product with an X suffix, as RX or SX, which reads like the extension codes KX and JX. The conductors are copper and a copper-nickel alloy, not platinum.

The IEC names for the same cable are RCA, RCB, SCA and SCB. Same drum, two naming conventions, and only one of them says the alloys were substituted. The high temperature side of this is covered in our Type S, R and B guide.

Grades and error limits

Thermocouple grade is the wire rated to form the sensing junction itself and to sit in the process temperature. Extension grade is the same alloy pair held to a looser tolerance and rated for ambient service, typically 0 to 200 °C. It is cheaper because it never has to survive the hot end.

The limits published for the cable are not sensor accuracy figures. They state how much error the cable itself may add to the circuit. That is why IEC expresses them in microvolts rather than degrees.

Cable	Standard limit	Special limit	Rated band
KX	±2.2 °C	Not listed	0 to 200 °C
JX	±2.2 °C	±1.1 °C	0 to 200 °C
EX	±1.7 °C	Not listed	0 to 200 °C
TX	±1.0 °C	±0.5 °C	0 to 100 °C
NX	±2.2 °C	Not listed	0 to 200 °C
R and S compensating	±5 °C	Not listed	0 to 200 °C
B compensating	0 to −3.7 °C	Not listed	0 to 100 °C

Values as published in a manufacturer limits-of-error sheet to the ANSI convention. Note the Type B row: the band is one-sided rather than symmetrical, so the cable can only pull the reading one way.

Two things in that table are easy to miss. The TX band stops at 100 °C rather than 200 °C. A Type T head in a hot cabinet can therefore be outside its cable rating while a Type K head beside it is inside. The R and S compensating figure of ±5 °C is also several times looser than any base metal extension cable, which is the price of not running platinum down the tray.

Type K has one more wrinkle of its own. KCA and KCB are different compensating alloys with different ceilings, and KCB was known as Type V for years. The choice between KX and the KC pair is set out in the Type K guide.

Why copper fails

A thermocouple does not measure the temperature of its junction. It measures the difference between the sensing junction and the reference junction. The instrument supplies that reference by measuring its own terminals and adding the matching voltage back in.

Splice copper into the run and you move the reference junction to the splice, because that is now where the alloy pair ends. The instrument keeps compensating for its own terminals, which are somewhere else and at a different temperature. The error is the voltage between those two places.

Error = E(T at the splice) − E(T at the terminals)

Type K, splice in a head at 100 °C, instrument terminals at 25 °C: 4.096 mV − 1.000 mV = 3.096 mV, which at 40.96 µV per °C is about 76 °C of error.

Both voltages are read from the NIST reference table for Type K. The sensitivity figure is the average slope of that same table between 0 and 100 °C. The direction follows from the sign. A splice hotter than the instrument terminals reads low; a splice colder than the terminals reads high.

That is the calculation behind the advice never to extend a thermocouple with copper. It also explains why the error is usually quoted as a vague number. It is not a property of the copper. It is the temperature difference between two points that vary from one installation to the next.

Resistance by wire gauge

Thermocouple alloys are poor conductors compared with copper, so a run of extension cable has real resistance. Suppliers publish it as ohms per double foot, meaning one foot of cable counting both conductors, measured at 20 °C.

AWG	Type K	Type J	Type T	Type E
14	0.147	0.089	0.074	0.176
16	0.232	0.141	0.117	0.277
20	0.588	0.357	0.297	0.702
24	1.488	0.905	0.754	1.778
28	3.590	2.297	1.920	4.330

Ohms per double foot at 20 °C, from published wire tables. A second manufacturer table agrees with these figures to within about three percent. Type E is the most resistive of the four and Type T the least.

What that resistance does depends entirely on the instrument. A modern thermocouple input is a voltage measurement with a very high input impedance, 78 MΩ on one sixteen channel module. Against that, the cable resistance does not divide the signal in any way you could measure.

What it does instead is convert the small bias current the input draws into an offset voltage. The same module specifies an added error of 0.05 µV per ohm once the source resistance passes 50 Ω. It also states an input current of 50 nA. Those two numbers are the same statement: 50 nA flowing through 1 Ω produces 0.05 µV.

100 ft of 24 AWG KX: 1.488 × 100 = 148.8 Ω. Added error = 0.05 × (148.8 − 50) = 4.94 µV, or 0.12 °C on a Type K.

A tenth of a degree over a hundred feet of thin cable is small beside the ±2.2 °C the cable itself is allowed. The number matters because it shows where the real limit sits, and it is not signal loss.

Run length by gauge

The limit that stops a long run is the allowable source resistance printed in the instrument specification. One temperature input module allows a maximum of 250 Ω for thermocouple and millivolt inputs. Divide that by the resistance per double foot and you have the answer in feet.

AWG	Type K, feet	Type K, meters	Type J, feet
14	1,700	518	2,809
16	1,078	328	1,773
20	425	130	700
24	168	51	276
28	70	21	109

Calculated as 250 Ω divided by the resistance per double foot in the table above. Check the figure in your own instrument manual before using these lengths. Allowable source resistance varies between makes, and some inputs accept far more than 250 Ω.

The shape of the table is the useful part. Going from 24 AWG to 20 AWG buys a factor of 2.5 in length, and going to 14 AWG buys a factor of ten. A run that fails on thin cable may succeed on the next size up with no change to the instrument.

One caveat belongs here. In an electrically noisy plant it is induced noise, not resistance, that ends a long run first, and no gauge table will show where that limit sits. Resistance is simply the part you can calculate in advance.

Insulation and jacket

The 0 to 200 °C rating on extension grade is a statement about the alloy tolerance, not about the plastic around it. The insulation carries its own limit, and the lower of the two governs the installation.

Insulation	Temperature range
PVC	−40 to 105 °C
FEP	−200 to 200 °C
PFA	−267 to 260 °C
Glass braid	−73 to 482 °C
Ceramic fiber braid	−73 to 1,204 °C

Ranges as published in a manufacturer insulation table. A high temperature braid raises the insulation limit, but on extension grade the wire tolerance still caps the assembly at 200 °C.

PVC is the common choice on stock extension cable, and it is the part that fails first. A head on a steam line, or a cable tray in direct sun in a hot climate, will pass 105 °C with no sign at the display. Nobody notices until the insulation softens and the pair shorts.

Polarity and wiring errors

A reversed pair is a classic wiring fault on a thermocouple circuit, and one of the hardest to spot. The display shows a plausible number rather than an alarm. The arithmetic shows what to look for.

Type K, process at 60 °C, terminals at 25 °C. Correct signal: 2.436 − 1.000 = 1.436 mV. Reversed, the instrument sees −1.436 mV and still adds its own 1.000 mV of compensation, giving −0.436 mV.

The reading therefore sits below the terminal temperature, in this case below zero, while the process is at 60 °C. Worse, it moves the wrong way. Heat the process and the displayed value falls, which is the signature to watch for during commissioning.

The size of the error is roughly twice the difference between the process and the terminals, so it grows as the plant warms up. On the same Type K at a process temperature of 300 °C the input would be driven to about −10.2 mV. That is far below anything the scale covers, so the instrument shows under-range or a sensor fault instead of a number.

Getting polarity right starts with reading the colors correctly, and the colors mean different things depending on which standard the cable was made to. Our thermocouple color code guide puts the ANSI, IEC and legacy charts side by side. It includes the two cases where a color means the opposite polarity from the one an engineer expects.

Two mechanical points matter as much as the colors. Both legs must cross from sensor to cable at the same place and at the same temperature. That is what a terminal block inside a head is for. Every intermediate terminal, plug or barrier strip is another place the pair can be crossed, which is the argument for fewer joints rather than more.

Ordering and installation

Six lines on a purchase order settle almost every cable argument: thermocouple type; grade, extension or compensating; the color code standard; insulation; conductor gauge; and whether the cable is screened. The gauge line is the one that sets the length you can run.

On installation, two rules carry most of the weight. Ground the screen at one end only, normally the instrument end, so it cannot carry a circulating current. Keep the run away from power cabling. Both apply to any low level signal, and they are covered for current loops in our 4-20 mA loop guide.

There is also a way to avoid the question entirely. Mount a temperature transmitter in the head and the long run becomes an ordinary copper pair carrying 4-20 mA. No extension cable, no compensating limits, and no polarity colors to decode at the far end. On a run of any length it removes the cable grade question entirely.

FAQ

Can you extend thermocouple wires?

Yes, with extension or compensating cable of the same thermocouple type, joined with the correct polarity at a single point. What you cannot do is splice in ordinary copper. That moves the reference junction to the splice and produces an error equal to the temperature difference between the splice and the instrument terminals.

What is the difference between thermocouple wire and extension wire?

Thermocouple grade wire is rated to form the sensing junction and to sit at process temperature. Extension grade uses the same alloy pair, but is held to a looser tolerance and rated only for ambient service, typically 0 to 200 °C. Extension grade is cheaper and belongs between the head and the instrument, never at the hot end.

What is a thermocouple extension cable?

It is cable made from the same alloy pair as the thermocouple it serves. The designation carries an X after the type letter, such as KX or JX. Because the alloys match, no new thermoelectric junction is created where it joins the sensor leads. The reference junction stays at the instrument, where compensation is applied.

Does the length of thermocouple wire matter?

It matters through resistance rather than signal loss. A hundred feet of 24 AWG Type K extension cable measures about 149 Ω, which on a high impedance input adds roughly 0.12 °C. The practical limit is the allowable source resistance in the instrument specification, 250 Ω on the module quoted in this guide, and heavier gauge buys length in direct proportion.

Request a quote

Tell us the application and we configure one assembly, not a shelf part. Give us the thermocouple type, the run length, the head temperature and the instrument you are wiring into. We specify the cable grade and gauge along with the sensor. Reach our application engineers or use the form below.

Written and technically reviewed by Wu Peng and the Instranova engineering team.