You cannot trim a thermocouple the way you can adjust an RTD, because a thermocouple is just two dissimilar wires that produce a voltage set by physics; there is nothing in the junction to calibrate. So calibrating a thermocouple loop really means two things: verifying that the reading device converts the thermocouple voltage into the right temperature, and confirming that its cold-junction compensation is correct. You source a precisely known millivolt for the thermocouple type, check that the transmitter reports the right temperature, and separately validate the actual junction in a dry block. Along the way the two things most likely to bite you are cold-junction errors and extension-wire polarity mistakes, both of which produce plausible but wrong readings.
Calibrate a Thermocouple in one line: Because a thermocouple junction cannot be adjusted, calibration means verifying the reading device and its cold-junction compensation. Source a known millivolt for the thermocouple type, such as type K, J, or T, with the correct reference-junction correction applied, and confirm the transmitter converts it to the right temperature. Then validate the physical junction by immersing the actual thermocouple in a dry-block calibrator at known temperatures. Watch for cold-junction compensation error and extension-wire polarity mistakes, both of which produce believable but incorrect readings, and check the thermocouple against its decalibration limits since junctions drift with exposure.
A thermocouple produces a small voltage that depends on the temperature difference between its measuring junction and its reference junction, and the voltage-to-temperature relationship is fixed for each type, K, J, T, E, N, and the rest. To test the reading device you disconnect the thermocouple and source a precise millivolt corresponding to a known temperature for that type, using a thermocouple source or documenting calibrator. If you source the millivolt that type K produces at 300 C and the transmitter reports 300 C, the conversion is correct; if it reports something else, the reading device is at fault, since the sourced voltage was exactly right. This isolates the electronics cleanly, the same way injecting resistance isolates an RTD transmitter.
The catch that makes thermocouple work different is cold-junction compensation. The published millivolt tables assume the reference junction sits at 0 C, but in a real transmitter the reference junction is the terminal block, which is at ambient. So the transmitter measures its own terminal temperature and adds the corresponding voltage back in before converting, which is cold-junction compensation. When you source millivolts, you must account for this: either use a calibrator that applies the correct reference-junction correction to match the transmitter's compensation, or the numbers will be offset by the terminal temperature. A calibrator set to simulate the thermocouple with its internal CJC handling matched to the transmitter is what makes the sourced value meaningful.
Checking the cold-junction compensation itself is a distinct step worth doing deliberately, because a CJC error biases every reading by the same amount regardless of the process temperature. One practical check is to source the millivolt for a known temperature with the reference-junction correction enabled and confirm the reported temperature is right; a consistent offset that tracks the terminal temperature points at a CJC problem rather than a span or conversion error. Because the CJC error is an offset tied to ambient, it is easy to mistake for a general calibration shift, so isolating it, and confirming the transmitter is reading its own terminal temperature correctly, saves chasing a phantom span error.
Extension and compensating wire is where field thermocouple loops most often go wrong, and the error is usually polarity. Thermocouple extension wire must match the thermocouple type and must be connected with the correct polarity all the way from the junction to the reference junction, because any junction of dissimilar metals along the way generates its own voltage. If the extension wire is reversed, or if the wrong type is used, additional thermoelectric junctions appear at the connections and inject an error that depends on the temperature at those connection points. The insidious part is that a reversed pair still produces a plausible temperature, just a wrong one, so nothing looks obviously broken.
Because the extension-wire error lives in the field wiring rather than the transmitter, it will not show up when you source millivolts at the transmitter terminals, which is exactly why validating the whole loop separately matters. Sourcing at the transmitter tests the electronics and CJC; it says nothing about the wire between the transmitter and the sensor. To catch a polarity or wire-type mistake you have to inject or measure at the field junction, or compare the loop reading against a reference sensor placed at the actual measuring point, so the extension run is inside the test. A loop that reads correctly with millivolts sourced at the head but wrong with the real junction at a known temperature has a wiring or sensor problem, and extension polarity is a prime suspect.
Validating the physical junction is done in a dry-block calibrator or a stirred bath, the same as for an RTD. You immerse the actual thermocouple, wired through its extension to its transmitter, and set the block to known temperatures, then read the loop output. This is the only step that tests the junction itself, and it is where decalibration shows up. Thermocouples drift with exposure: prolonged high temperature, oxidation, and contamination gradually change the junction's output, and each type has practical limits beyond which its reading is no longer reliable. A junction that reads correctly at the head with sourced millivolts but wrong in the dry block has drifted and needs replacement, since there is no trim to recover it.
A thermocouple that cannot be trimmed is a thermocouple you must watch, because when it drifts the only remedy is replacement and the only warning is the reading itself moving away from truth. That makes continuous monitoring especially valuable for thermocouple points, since the slow decalibration of a junction shows up as a gradual bias in the trended temperature long before a scheduled check would catch it. Rather than trusting a junction blindly until it fails, you can see it start to wander.
When a cloud SCADA platform such as Merobix trends thermocouple temperatures continuously, several thermocouple-specific failures become visible in the data. A gradual downward or upward creep suggests a drifting junction near the end of its life; a sudden step suggests a wiring change or a new junction introduced by a repair; a bias that varies with ambient temperature at the terminal points suggests a cold-junction compensation problem. Seeing which pattern the trend shows tells a technician what to check, whether that is the junction in a dry block, the extension wiring, or the transmitter's CJC.
The trend also confirms a repair the way the dry block confirmed the diagnosis. After replacing a decalibrated junction or correcting reversed extension wire, the recorded temperature should return to agreement with neighboring measurements or with a reference, and a formerly ambient-linked bias should disappear once a CJC fault is fixed. Because the history from before the work is retained, you get an independent before-and-after that proves the replacement or rewiring actually corrected the reading, which is reassuring precisely because a thermocouple gives you no adjustment to fall back on if the fix was aimed at the wrong cause.
Not in the sense of trimming it, because a thermocouple is two dissimilar wires whose voltage is fixed by physics with nothing to adjust. Calibrating a thermocouple loop instead means verifying the reading device and its cold-junction compensation by sourcing a known millivolt, and validating the physical junction in a dry block. If the junction has drifted beyond its limits, the only remedy is replacement, since there is no trim to bring it back.
A thermocouple's output depends on the temperature difference between its measuring junction and its reference junction, and the standard voltage tables assume the reference is at 0 C. Since the real reference junction is the terminal block at ambient, the transmitter measures its terminal temperature and adds the matching voltage back in, which is cold-junction compensation. You check it by sourcing a known millivolt with the reference-junction correction enabled and confirming the reported temperature is right, watching for a consistent offset that tracks the terminal temperature.
Thermocouple extension wire must match the sensor type and keep correct polarity along the whole run, because any junction of dissimilar metals generates its own voltage. Reversed or wrong-type extension wire adds unwanted thermoelectric junctions at the connections, injecting an error that depends on the temperature there. The trouble is that a reversed pair still produces a plausible temperature, just an incorrect one, so it will not show up when you source millivolts at the transmitter and only appears when you test the whole loop through the real junction.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.