Most “bad thermocouple” reports are installation problems. The sensor reports the temperature of its junction — the installer's job is to make the junction's temperature equal the process temperature, and to deliver the millivolt signal to the instrument uncorrupted.
How deep should a Type K thermocouple be immersed?
Immerse at least 10–20× the sheath diameter (plus the sensitive tip length). Shallower immersion lets heat conduct out of the measuring zone along the stem, biasing the reading toward ambient — the single most common installation error.
Gas service needs the deeper end of that range; liquids, with their higher film coefficients, tolerate the shallower end. Where geometry forbids adequate depth, run the lead-in along an isotherm inside the process before exiting, so the conduction path sees no gradient.
Which junction style fits the duty?
Grounded junctions respond fastest but give no electrical isolation; ungrounded junctions isolate at the cost of response time; exposed junctions are fastest of all but only for benign, dry, non-pressurized media.
With grounded junctions, keep exactly one ground point in the measuring circuit — a second ground (often the instrument input) creates a loop that injects noise and offset, the classic problem near motors and drives. Isolated instrument inputs or ungrounded junctions break the loop.
Wire gauge sets the real temperature ceiling
The 1372 °C figure is the range of the reference function, not a duty rating. ASTM E230 Table 6 recommends upper limits for protected tube-and-wire Type K elements that fall with wire size — from 1260 °C at 8 AWG down to 760 °C at 30 AWG.
| Wire size | Diameter | Upper limit |
|---|---|---|
| 8 AWG | 3.25 mm | 1260 °C |
| 14 AWG | 1.63 mm | 1090 °C |
| 20 AWG | 0.81 mm | 980 °C |
| 24 AWG | 0.51 mm | 870 °C |
| 28 AWG | 0.33 mm | 870 °C |
| 30 AWG | 0.25 mm | 760 °C |
These limits apply to conventional tube-and-wire construction in closed-end protection tubes; they do not apply to compacted mineral-insulated (MI) sheathed cable, which is rated by sheath alloy and diameter instead. See MI Type K probes.
Extension wire and the signal path
Use genuine Type KX extension or the correct compensating grade, with correct polarity through every connector and terminal block. Any mismatched or reversed segment is itself a thermocouple pair and injects its own EMF.
Carry the color code through the run — KP yellow / KN red under ANSI MC96.1, KP green / KN white under IEC 60584-3 — and use Type K-bodied connectors, not copper terminals, anywhere a temperature gradient can exist across the connection. For MI cable, respect the minimum bend radius from the cable datasheet (commonly specified around 3× the sheath diameter) and avoid work-hardening the run with repeated re-forming, which shifts EMF locally.
Control the cold junction
Keep the reference junction isothermal and away from gradients, hot motors, and direct sun. A drifting terminal-block temperature shows up as a drifting process reading and is routinely misdiagnosed as sensor drift.
If the instrument's CJC sensor is remote from the actual termination, the difference between the two is a direct measurement error. Verify with the converter: enter the measured mV with the true terminal temperature and compare against the loop's displayed value. Background on the math is in the cold-junction compensation notes.
Thermowells and vibration
A thermowell protects the element but adds lag and a vibration failure mode: vortex shedding at the well's natural frequency fatigues the assembly. Check the wake-frequency margin against process velocity before installation, not after the first failure.
Spring-loaded elements keep the tip seated against the well bottom for thermal contact and damp relative motion. Sizing and materials are covered under thermowell selection; for the sensor build itself — sheath alloy, diameter, junction, certification — start from the Type K product family or the quote form.