What the Thermocouple Temperature Converter does
This converter turns thermocouple millivolts into temperature and back, for types K, J, T, E, N, R, S and B, using the ITS-90 reference functions and inverse polynomials published by NIST. It applies cold-junction compensation, shows the IEC 60584-1 tolerance class limits at the temperature you are working at, and prints a reference table for the type.
A thermocouple does not measure temperature. It measures the difference between the hot junction and wherever its wires end - the cold junction - so the millivolts on your meter already have the room subtracted from them. That is why a reading that looks 20 or 25 degrees low is usually not a faulty sensor but a missing compensation step, and why this page asks for the terminal temperature before it will answer.
Everything runs in the browser against a coefficient file served from this site. No readings are sent anywhere.
How to use it
- Pick the thermocouple type. The usable range and EMF limits for that type appear underneath; going outside them is refused rather than extrapolated.
- Enter the cold-junction temperature: the temperature of the instrument terminals or the reference block. Use 0 only if the instrument has already compensated, or if the junction really is in an ice bath.
- Choose the direction and type the value. Millivolts to temperature is the field case; temperature to millivolts is what you need when simulating a sensor with a calibrator.
- Read the tolerance table to see what the standard permits for each class at that temperature - it is usually far larger than people expect, and much larger than the conversion's own error.
- Open the reference table for a printable EMF-against-temperature list at whatever step you choose, and export it as CSV.
Reading the results
The hot-junction temperature is the answer. The EMF against 0 C is the tabulated value for that temperature; the cold-junction EMF is what the terminals contribute; the instrument reading is the difference between the two.
The inverse polynomial's error band is quoted per subrange, typically a few hundredths of a degree. Compare it with the class limits below: a class 2 type K at 800 C is permitted to be 6 degrees out, so the arithmetic is never what limits the measurement.
A tolerance class describes a new thermocouple. Drift in service, particularly in type K above about 500 C, is not covered by it and is the reason thermocouples are replaced rather than recalibrated.
If a reading is out by roughly the room temperature, check compensation before anything else. The page shows what the same millivolts would mean uncompensated so the size of the error is obvious.
Worked example: a type K reading 11.209 mV with the terminals at 25 C
A type K couple on a furnace wall reads 11.209 mV. The instrument's terminal block is at 25 C.
From the ITS-90 reference function, a type K at 25 C generates 1.000 mV against a 0 C reference. The thermocouple only produced the difference, so the true EMF against 0 C is 11.209 + 1.000 = 12.209 mV.
Inverting 12.209 mV gives 300.0 C. That is the hot junction.
Had the same 11.209 mV been read as if the cold junction were at 0 C, the answer would have been 275.8 C - 24 C low, which is roughly the terminal temperature. The NIST table brackets it: 275 C is 11.176 mV and 276 C is 11.217 mV.
At 300 C a class 1 type K is permitted to be the greater of 1.5 C and 0.4% of 300 C, so 1.5 C; a class 2 is permitted 2.5 C. The conversion's own inverse error over this subrange is between -0.05 and 0.04 C, which is thirty times smaller than the sensor's allowance.
Formulas and scoring rules
- EMF from temperature
E = sum(c_i x t^i) for the subrange containing tITS-90 reference function, EMF in mV against a 0 C reference junction.- Type K exponential term
E = polynomial + a0 x exp(a1 x (t - a2)^2) for t > 0 COnly type K carries this extra term; a0 = 0.1185976, a1 = -1.183432e-4, a2 = 126.9686.- Temperature from EMF
t = sum(d_i x E^i) for the subrange containing ENIST's approximate inverse. Its error band per subrange is shown with every answer.- Cold-junction compensation
E(against 0 C) = E(measured) + E(cold junction temperature)Then invert. Going the other way, the instrument reads E(hot) - E(cold).- IEC 60584-1 class limit
limit = max(absolute limit, fraction x |t|)For example type K class 1: max(1.5 C, 0.004 x |t|). Types R and S class 1 above 1100 C use 1 + 0.003(t - 1100) C.
Choosing a type
Type K is the default for general industrial work: a wide range, roughly 41 microvolts per degree, cheap. It drifts above about 500 C in the presence of oxygen and it is not linear, particularly near 0 C. Type N was designed to fix exactly that drift and is the better choice for long-term high-temperature work.
Type J is common on older plant and plastics machinery but the iron leg rusts, and it stops at 750 C. Type T is the one for cold work down to -200 C and is very stable but tops out at 400 C. Type E has the largest output of the standard types, which makes it good where the signal has a long run to travel.
The noble-metal types R, S and B are for furnaces and kilns. They reach 1768 C (R and S) or 1820 C (B) but produce barely a tenth of type K's output, so the instrument has to be much better. Type B is unusual in producing almost nothing below 250 C - it cannot be inverted at all there - which is why it does not need cold-junction compensation below about 50 C.
What the millivolts do not tell you
Extension and compensating cable must match the thermocouple type and its polarity. A reversed junction in a terminal box produces a reading that moves the wrong way, and a copper joint anywhere in the run creates a new, unintended cold junction at the temperature of that box.
Grounded, ungrounded and exposed junctions differ in response time and in how much common-mode noise they let through. None of that changes the millivolt-to-temperature relationship, but all of it changes what the millivolts are.
Limitations: what the result does not prove
- It converts EMF to temperature. It cannot tell you whether the thermocouple is intact, correctly installed, in the right place, or wired with the right extension cable - all of which change the reading far more than the conversion does.
- Tolerance classes apply to a new thermocouple within the range stated for that class. Drift in service, especially type K above 500 C, is not covered.
- The inverse polynomials are approximations with a published error band per subrange, shown with each answer. The direct reference functions are the definition and carry no such error.
- Sensors sold as type K are not always to IEC 60584-1. Special-limits and non-standard alloys exist; the calibration certificate, if there is one, takes precedence over any table.
Privacy: where your data goes
Everything you paste, type or drop is processed in this browser tab. It is not uploaded, logged, stored or sent to analytics. Session recording and tag-manager scripts are switched off on this page.
Standards and sources
- IEC 60584-1 - Thermocouples: EMF specifications and tolerances - checked 19 Sep 2026
- NIST ITS-90 Thermocouple Database (SRD 60)
- NIST Monograph 175 - Temperature-Electromotive Force Reference Functions and Tables
Frequently asked questions
How do I convert type K millivolts to temperature?
Add the EMF that your cold-junction temperature would itself produce, then invert the total with the ITS-90 polynomial for type K. This page does both: enter the meter reading and the terminal temperature, and it returns the hot-junction temperature along with the EMF against a 0 C reference.
What is cold-junction compensation and why does it matter?
A thermocouple generates a voltage from the difference between its hot and cold junctions, so the meter sees E(hot) minus E(cold). Compensation adds the cold-junction EMF back before converting. Skip it and the answer is low by roughly the terminal temperature - about 25 C in a normal control room.
How accurate is a type K thermocouple?
IEC 60584-1 class 1 permits the greater of 1.5 C and 0.4% of the reading, class 2 the greater of 2.5 C and 0.75%. At 800 C that is 3.2 C and 6.0 C respectively, for a new sensor. Drift in service is additional and is not covered by the class.
Why does type B have no reading below 250 C?
Type B produces almost no output at low temperatures - its EMF curve doubles back near 42 C - so a single millivolt value there can mean two different temperatures. NIST's inverse polynomial therefore starts at 250 C, and this page refuses to invert below it rather than picking one of the two answers.
Which thermocouple type should I use above 1200 C?
Type N reaches 1300 C and type K nominally 1372 C, but both drift badly at that end. For sustained work above 1200 C the noble-metal types are the practical answer: R and S to 1768 C, B to 1820 C, at the cost of about a tenth of the output and a much more expensive sensor.
Can I use type K extension wire on a type J thermocouple?
No. Extension cable has to match the thermocouple alloys, because any junction between dissimilar metals generates its own EMF. Mixing types puts an uncontrolled voltage in series with the measurement that changes whenever the terminal box changes temperature.
What EMF does a type K produce at 100 C?
4.096 mV against a 0 C reference junction, per the ITS-90 table. If the instrument terminals are at 25 C it will only see 4.096 - 1.000 = 3.096 mV, which is exactly why compensation exists.
Where do these coefficients come from?
The NIST ITS-90 Thermocouple Database (Standard Reference Database 60, and NIST Monograph 175), which is the source IEC 60584-1 adopts for the reference functions. The coefficient file on this site records its source and the date it was checked.
Last reviewed by the A2Z.Tools team against the sources listed above.
