Thermocouple Voltage Calculator

Thermocouple output example: a Type K thermocouple at 100°C (with a 0°C cold junction reference) outputs roughly 4.1 mV — small enough that cold junction compensation is essential for an accurate reading. A thermocouple is two dissimilar metal wires joined at one end (the measuring or "hot" junction). When that junction is at a different temperature from the point where the wires connect to your measuring instrument (the reference or "cold" junction), the Seebeck effect generates a small DC voltage between the two free ends — typically a few tens of microvolts per degree of temperature difference. This calculator converts junction temperatures directly into that output voltage, and back-calculates what your instrument actually measures once cold junction compensation is accounted for.

Formula used: the calculator evaluates E(T), the NIST ITS-90 reference EMF function for the selected type, at both the hot and cold junction temperatures, then takes the difference: Eₙₜ = E(Tₕ) − E(T). E(T) itself is not a single simple equation — it is a piecewise polynomial in temperature, with a different set of coefficients for sub-zero and above-zero ranges (and, for Type K, an additional small exponential correction term above 0°C), fitted by NIST to match physically measured reference thermocouple data as closely as possible across each type's full rated range.

Why subtraction, not just E(Tₕ) alone: the published NIST tables for E(T) are all built on the assumption that the reference junction sits at exactly 0°C (the ice point). In practice, almost no real instrument maintains a genuine ice-point reference — instead, it measures the actual temperature of its own cold junction (often the terminal block where the thermocouple wire connects to copper wiring) using a separate temperature sensor, and performs this same subtraction internally. This process is called cold junction compensation, and getting it wrong is one of the single biggest sources of error in thermocouple-based temperature measurement.

Worked example: a Type K thermocouple with its hot junction at 500°C and its cold junction at a typical ambient 25°C: E(500) ≈ 20.644 mV, E(25) ≈ 1.000 mV, so net output ≈ 19.644 mV. If you incorrectly assumed a perfect 0°C cold junction instead of the real 25°C ambient, you would read E(500) ≈ 20.644 mV directly — an error of roughly 1 mV, which for Type K corresponds to about 24°C of apparent temperature error. This is exactly why cold junction compensation matters even for a "small" 25°C reference offset.

The Seebeck coefficient is not constant: a common simplification treats thermocouple output as strictly linear (voltage = constant × temperature difference), but the real relationship is mildly non-linear across the full range of any type — the local slope of E(T), called the Seebeck coefficient at that temperature, changes gradually as temperature increases. Near room temperature Type K sits around 40-41 µV/°C, but it rises to roughly 42 µV/°C by a few hundred degrees Celsius before the curve's shape changes again at higher temperatures. This is precisely why NIST publishes tables (and the underlying polynomials this calculator evaluates) rather than a single linear conversion factor — a linear approximation is fine for a rough estimate but introduces meaningful error across a wide operating range.

Where the coefficients come from: NIST derived these reference functions from an extensive set of calibrated laboratory measurements against the International Temperature Scale of 1990 (ITS-90), then fit high-order polynomials (piecewise by range) that reproduce those measurements to within a very small fraction of a microvolt across most of each type's range. These are the same reference functions built into essentially every commercial temperature transmitter, PLC analog input card, and calibration standard that claims ITS-90 compliance for thermocouple linearization.

Reading the result the other way (temperature from voltage): everything above works forward — from temperature to voltage. Real instruments more often need the reverse: they measure a voltage and need to report a temperature. NIST also publishes separate inverse polynomials (voltage-to-temperature) for exactly this purpose; they are mathematically distinct fits, not simply the algebraic inverse of the forward function, chosen specifically to minimize error when solving in that direction. This calculator implements the forward (temperature-to-voltage) direction, which is the more useful direction for checking expected sensor output, verifying a calibrator's source setting, or sanity-checking a data sheet figure.

Extension and compensating cable: the wire that runs from the thermocouple junction back to the instrument must itself be made from matching (or closely matched, "compensating") alloy pairs for each type — using ordinary copper wire, or the wrong type's extension cable, effectively creates additional unintended thermocouple junctions at every connection point, each generating its own spurious EMF that corrupts the reading. This is a wiring and installation issue entirely separate from the voltage calculation itself, but it is worth understanding because it is one of the most common real-world causes of a "correct calculation, wrong reading" symptom.

Polarity matters: each thermocouple type has a designated positive and negative leg (for Type K, the positive leg is magnetic — attracted to a magnet — while the negative leg is not), and reversing the two at any connection point in the circuit, including at the instrument terminals, effectively subtracts where it should add, producing either a wildly wrong reading or, at certain temperature combinations, a near-zero or negative reading that can look like a sensor fault rather than a wiring error.

Accuracy classes: IEC 60584 and ASTM E230 define standard and "special limits of error" accuracy classes for each thermocouple type, expressed as either a percentage of reading or a fixed degree tolerance, whichever is larger, over specified temperature ranges. This calculator reports the theoretical, error-free EMF from the reference function; real sensor tolerance, wiring resistance effects, and instrument input accuracy all add to that baseline and should be considered separately when specifying overall measurement uncertainty for a system.

Isothermal blocks and multiple cold junctions: in a real instrument with multiple thermocouple input channels, all channels typically terminate on a single isothermal (uniform-temperature) block, with one shared cold-junction temperature sensor reading that block's actual temperature. If that block is not genuinely isothermal — for example, one corner sits closer to a heat-generating component inside the instrument enclosure — different channels effectively see slightly different, unaccounted cold-junction temperatures, introducing small but real channel-to-channel measurement offsets that no amount of correct calculation elsewhere in the circuit can fix.

Practical use of this calculator: use it to predict expected sensor output before wiring up a new installation, to verify a signal calibrator or simulator's mV output setting against a target temperature, to sanity-check a data sheet or datalogger reading against hand-calculated expectations, or to understand how much a given cold-junction temperature error would shift your apparent reading — all without needing to look up and manually interpolate a printed NIST reference table.

Summary: net thermocouple output equals E(Tₕ) minus E(T), evaluated using the ITS-90 reference polynomial for the selected type; cold junction temperature must always be accounted for; the relationship is mildly non-linear across each type's range, which is exactly why a polynomial fit (rather than a single conversion constant) is used; and correct wiring polarity and matching extension cable are separate but equally important prerequisites for the calculated figure to match what your instrument actually reads.

1. Ignoring cold junction temperature entirely. Reading E(Tₕ) alone and treating it as the expected instrument output assumes a perfect 0°C reference — for a realistic 20-30°C ambient cold junction, this alone can introduce tens of degrees of apparent error depending on the type.

2. Using ordinary copper wire instead of matched extension cable. Every unintended junction formed by mismatched wire creates its own spurious EMF at its own local temperature, corrupting the signal in a way that is often intermittent and hard to diagnose because it depends on ambient conditions along the run.

3. Reversed polarity at any connection point. One swapped connection, even a single terminal deep in an otherwise correct run, can flip the sign of part of the signal path and produce a reading that looks plausible but is systematically wrong.

4. Assuming a linear mV-per-degree conversion across a wide range. The Seebeck coefficient changes gradually with temperature for every type — a linear estimate anchored at one point on the curve drifts increasingly out of true the further you move from that anchor point.

5. Operating a type outside its recommended atmosphere. Type J's iron leg oxidizes in air at sustained high temperature; Type K can suffer "green rot" (preferential oxidation) in certain marginally reducing/oxidizing mixed atmospheres — both effects shift the actual EMF-vs-temperature relationship away from the ideal reference function over time, independent of any calculation error.

6. Mixing up ANSI and IEC/BS color codes. The same thermocouple type uses different colored insulation depending on the standard the manufacturer follows — assuming a color code without checking the specific standard in use can lead to a correctly-calculated but physically miswired circuit.

7. Not accounting for an isothermal block that isn't actually isothermal. Multiple channels sharing one cold-junction sensor assume every channel's physical termination point is at the same temperature as that sensor — a poorly designed or thermally uneven terminal block silently introduces channel-specific offsets.

8. Treating calculated EMF as the full measurement uncertainty. This calculator's output is the theoretical, error-free reference value — real installation tolerance stacks on top from sensor manufacturing tolerance (per IEC 60584/ASTM E230 accuracy class), wiring, and instrument input accuracy, and should be added separately when specifying total system uncertainty.