A temperature transmitter has two jobs stacked on top of each other: convert a sensor signal into a temperature, and convert that temperature into a 4-20 mA output. Calibrating it well means checking both jobs, ideally in two stages, so you know exactly where any error lives. The clean approach is to inject a precisely known sensor signal, an RTD resistance or a thermocouple millivolt, to test the transmitter electronics alone, then optionally put the real sensor and transmitter together into a dry block to verify the whole chain end to end. Splitting the work this way keeps a drifting sensor from being blamed on the transmitter, and vice versa.
Calibrate a Temperature Transmitter in one line: To calibrate a temperature transmitter, first inject a known RTD resistance or thermocouple millivolt with a simulator and confirm the transmitter reports the corresponding temperature and drives the correct 4-20 mA output, adjusting the input trim or output trim as needed. Then, if you want to prove the sensor too, immerse the actual sensor and transmitter in a dry-block calibrator at known temperatures and check the whole loop. Sourcing thermocouple millivolts requires correct cold-junction compensation, and RTD work benefits from sensor matching when the transmitter supports it.
The first stage tests the transmitter as an electronic converter, with no real sensor involved. For an RTD input you disconnect the sensor and connect a decade resistance box or RTD simulator, then dial in resistances that correspond to known temperatures from the sensor table. Setting 100.00 ohm should make a Pt100 transmitter read 0 C; setting the table value for the top of your range should make it read the span endpoint. Because the injected resistance is exactly known, any difference between the temperature the transmitter reports and the temperature that resistance represents is transmitter error, cleanly isolated from the sensor.
For a thermocouple input the same idea applies, but you source a millivolt signal instead of a resistance. A thermocouple millivolt source or documenting calibrator lets you inject the exact voltage that the sensor type produces at a chosen temperature. The subtlety here is cold-junction compensation: a real thermocouple only produces the tabulated voltage relative to a reference junction, so the transmitter adds a correction for the temperature at its own terminals. When you source millivolts you must let the calibrator apply the matching reference-junction correction, or the transmitter will report a temperature offset by the terminal temperature. Getting the CJC handling right is the difference between a valid thermocouple calibration and a confusing one.
With the known signal injected, you compare the transmitter output at several points across the range, typically the bottom, middle, and top, and check both the digital temperature reading and the analog output. Two kinds of adjustment may be available. An input or sensor trim aligns the transmitter's interpretation of the incoming signal, correcting the temperature it derives from a given resistance or millivolt. An output trim aligns the 4-20 mA it produces for a given temperature. Knowing which one is off, the reading or the current, tells you which trim to touch.
Many smart transmitters support sensor matching, sometimes called Callendar-Van Dusen matching, which is worth understanding because it changes what calibration means. Instead of assuming the RTD follows the generic IEC 60751 curve, the transmitter is loaded with the specific coefficients from that individual sensor's calibration certificate. The transmitter then computes temperature from resistance using the real sensor's curve rather than the nominal one. When a matched sensor and transmitter are paired, the combined error can be markedly smaller than either the nominal sensor or a generic transmitter would give, which is why high-accuracy loops use it. On the bench, matching means the resistances you inject must correspond to that sensor's actual curve, not just the standard table.
The distinction between input trim and output trim is the concept people most often blur, and it matters because they fix different faults. An input trim, or sensor trim, corrects the front end: if a known 138.5 ohm makes the transmitter report 99.2 C instead of 100 C, the input trim brings that reading back to 100 C. An output trim, or digital-to-analog trim, corrects the back end: if the transmitter correctly reports 100 C but outputs 11.9 mA where the range calls for 12.0 mA, the output trim aligns the current. A transmitter can be perfect on one and off on the other, so a thorough calibration checks the reading against the injected signal and the current against a reference milliamp meter separately.
Deciding when to adjust versus when to leave alone follows the same as-found, as-left discipline used across instrument calibration. You first record the as-found error at each point without touching anything, because that record is what proves whether the loop was within tolerance while it was in service. Only if the as-found error exceeds the calibration tolerance do you trim, and then you record the as-left error to confirm the adjustment landed. Trimming a transmitter that was already inside tolerance adds no accuracy and destroys the as-found history that told you the loop had been trustworthy.
Stage one proves the transmitter but says nothing about the sensor, because you replaced the sensor with a simulator. To prove the entire measurement chain you run stage two: immerse the actual RTD or thermocouple, still wired to its transmitter, into a dry-block calibrator or stirred bath at known temperatures, and read the output. Now the error you see is the combined error of the sensor plus the transmitter, which is what the process actually experiences. A loop that passed stage one but fails stage two has a drifted or damaged sensor, and you have narrowed the fault without guessing.
This is where a monitoring platform earns its place, because the whole point of calibrating a temperature transmitter is that its output feeds a control or monitoring system that people trust. When a cloud SCADA system such as Merobix trends a temperature point continuously, a slow calibration drift shows up as a gradual bias in the recorded data long before a scheduled calibration would catch it. That lets you calibrate on evidence rather than only on a calendar, prioritizing the transmitters whose trends have started to wander over the ones that have held steady.
After calibration, the same trend confirms the work. A transmitter that was reading two degrees high and was trimmed back to reference should show that bias vanish in the historical data, and a formerly noisy point should quiet down if the fix addressed a bad connection. Because the as-found and as-left values are recorded on the calibration sheet and the live behavior is recorded in the SCADA history, you end up with an independent before-and-after that verifies the calibration actually improved the measurement the process depends on.
Input trim, also called sensor trim, corrects how the transmitter converts the incoming resistance or millivolt into a temperature reading. Output trim, or digital-to-analog trim, corrects the 4-20 mA current the transmitter produces for a given temperature. A transmitter can be accurate on one and off on the other, so you check the reading against the injected signal and the current against a reference meter separately, then trim only the stage that is out.
Not for the electronics. You can inject a known RTD resistance or thermocouple millivolt with a simulator and fully test and trim the transmitter without the real sensor connected. You only need the actual sensor when you want to verify the complete measurement chain, in which case you immerse the sensor and transmitter together in a dry block at known temperatures to capture the combined sensor-plus-transmitter error.
A thermocouple only produces its tabulated voltage relative to a reference junction, so the transmitter adds a correction based on the temperature at its own terminals. When you source millivolts from a calibrator you must enable the matching reference-junction correction, otherwise the transmitter applies its cold-junction compensation to a signal that did not account for it and reports a temperature offset by the terminal temperature. Getting the CJC handling consistent between source and transmitter is essential to a valid thermocouple calibration.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.