F
Member
【Editorial Discussion】
A thermocouple and an RTD (or two thermocouples) that agree at the bench and disagree after a long run to the cabinet are a classic instrumentation argument. Cold-junction compensation, the extension alloy, and interference on the cable can each produce a stable-looking offset. Swapping the sensor first is often the slow path.
1) Confirm the cold junction. A thermocouple input that uses the wrong CJC location, an uncalibrated terminal temperature, or a junction that is not actually at the terminals will read a consistent error that looks like “the process is off.” If the error tracks cabinet temperature more than process temperature, start here.
2) Extension wire must match the thermocouple type and polarity for the whole run. Copper, the wrong alloy, or an extra junction in a terminal strip creates new thermoelectric voltages. An RTD in four-wire or proper three-wire compensation is a different budget: its long-cable error is more about resistance imbalance and excitation, not thermoelectric EMF — do not apply the same correction story to both.
3) EMI and grounding show up as noise or as a rectified offset, especially near drives and welders. Shield termination, pair twisting, and whether the input is isolated should be checked before declaring the sensor out of tolerance. A handheld meter at the sensor head versus the cabinet input splits the run from the element.
4) Write down the measurement point, the cable type, and the input configuration next to the two readings. Without that, a “two-degree disagreement” is not a reproducible fact.
Open question: when your last long-run disagreement was resolved, was it CJC, the extension alloy, or interference? Do you standardize on bringing an RTD in for this reason, or do you make the thermocouple run work? Please share the check order you use.
A thermocouple and an RTD (or two thermocouples) that agree at the bench and disagree after a long run to the cabinet are a classic instrumentation argument. Cold-junction compensation, the extension alloy, and interference on the cable can each produce a stable-looking offset. Swapping the sensor first is often the slow path.
1) Confirm the cold junction. A thermocouple input that uses the wrong CJC location, an uncalibrated terminal temperature, or a junction that is not actually at the terminals will read a consistent error that looks like “the process is off.” If the error tracks cabinet temperature more than process temperature, start here.
2) Extension wire must match the thermocouple type and polarity for the whole run. Copper, the wrong alloy, or an extra junction in a terminal strip creates new thermoelectric voltages. An RTD in four-wire or proper three-wire compensation is a different budget: its long-cable error is more about resistance imbalance and excitation, not thermoelectric EMF — do not apply the same correction story to both.
3) EMI and grounding show up as noise or as a rectified offset, especially near drives and welders. Shield termination, pair twisting, and whether the input is isolated should be checked before declaring the sensor out of tolerance. A handheld meter at the sensor head versus the cabinet input splits the run from the element.
4) Write down the measurement point, the cable type, and the input configuration next to the two readings. Without that, a “two-degree disagreement” is not a reproducible fact.
Open question: when your last long-run disagreement was resolved, was it CJC, the extension alloy, or interference? Do you standardize on bringing an RTD in for this reason, or do you make the thermocouple run work? Please share the check order you use.