Thermocouple Loop Whose Reading Tracks Enclosure Temperature
A Type K thermocouple with a grounded measuring junction is installed on a reheat furnace controlled at 900 °C. Type K extension wire runs from the thermocouple head to a 4–20 mA transmitter mounted in a field enclosure on the platform; the transmitter's 24 V loop supply is shared with a variable-frequency-drive cabinet. At 07:00 the enclosure sits at 22 °C and the DCS indicates 902 °C. At 15:00 the enclosure reaches 46 °C and the DCS indicates 926 °C. A calibrated handheld meter connected at the thermocouple head reads 903 °C at both times. Maintenance records show the transmitter was replaced last month and its cold-junction setting was left at "external, 0 °C" because the internal sensor "looked noisy." Separately, whenever the VFD runs above 30 Hz, a 50 Hz ripple of about 0.3 mV appears on the current loop and the indicated temperature jumps by roughly 5 °C. The furnace set point has not changed and the furnace is known to hold ±3 °C.
1. Is the 24 °C span error a cold-junction fault or a sensor/process fault? Support the answer with numbers and give one decisive test that takes under ten minutes.
Read the worked answer
The error is 24 °C and it maps one-to-one onto the enclosure temperature (22 °C gives 902 °C, 46 °C gives 926 °C). Type K sensitivity near 900 °C is about 42 µV/°C, so an uncompensated junction at the transmitter terminals produces roughly 24 × 42 µV = 1.0 mV, exactly the size of error seen. With CJC set to "external, 0 °C" the transmitter adds no correction, so indicated temperature equals true temperature plus terminal-block temperature. The handheld reading of 903 °C at the head at both times rules out the sensor, the extension wiring and the process. Decisive test: short the transmitter input terminals with a copper link and read the indicated value. A working internal CJC reads the terminal-block temperature (about 46 °C at 15:00); the present configuration reads 0 °C. Equivalently, inject 37.3 mV (900 °C for Type K against a 0 °C reference) and see whether the display shows 900 or about 924 °C. Corrective action is to set CJC to internal, or fit a transmitter with integral cold junction and an isothermal terminal block.
2. Does the 50 Hz ripple share the same root cause? Explain the mechanism and propose a fix.
Read the worked answer
No. The signatures differ. The cold-junction error is DC, tracks enclosure temperature, and is identical whether the loop is scanned or read locally. The ripple is AC, appears only when the VFD exceeds 30 Hz, and 0.3 mV at 42 µV/°C is about 7 °C of apparent temperature, consistent with the reported 5 °C jumps. The mechanism is a ground loop: the grounded thermocouple ties the furnace shell to the transmitter input, and VFD return currents raise the shell potential relative to the transmitter's ground reference. The resulting common-mode voltage, working against the finite CMRR of the input, appears as a differential error, and a small circulating current can flow in the loop wiring. Confirm by clamping a current probe around the loop, or by temporarily powering the transmitter from an isolated supply and watching the ripple vanish. Fixes, in order: an isolated transmitter or isolated input card; single-point grounding of the loop supply; shielded twisted pair with the shield grounded at the receiving end only; and routing the loop cable away from VFD power cables. The 4–20 mA signal itself is not at fault.
3. A colleague suggests replacing the thermocouple with a three-wire Pt100 to remove the ambient problem. Assess this proposal.
Read the worked answer
It addresses the largest error but not the whole installation. An RTD has no reference junction, so the 24 °C ambient sensitivity disappears, and a Pt100 at 900 °C is practical only in a protection tube with a suitable sheath and a high-temperature element; the transmitter input must also be reconfigured for RTD and for three-wire lead compensation. Three-wire compensation assumes the two current-carrying leads are equal; a 0.2 Ω mismatch in 60 m of thin copper is worth about 0.5 °C, which is acceptable here. It does nothing for the VFD-induced ripple, which is a grounding problem set by the grounded junction and the shared supply; an ungrounded or isolated RTD circuit would actually reduce that path, but the loop grounding still needs correcting. Cost matters: an RTD at 900 °C is more fragile and slower than a thermocouple, and the existing head and extension wiring would change. The cheaper fix is to correct the CJC setting and isolate the input.