Thermocouple Accuracy & Sources of Error
A thermocouple's published tolerance class is only the starting point for real-world accuracy. Installation, cold-junction compensation, wiring, and drift all add error. This guide separates the initial tolerance from the in-service error budget and shows where accuracy is won or lost.
Initial Tolerance (the Starting Point)
ASTM E230, E1751, and E696 define Standard and Special tolerance classes — the limits of error for new wire. For Type K, for example, Standard tolerance is the greater of ±2.2 °C or ±0.75%, and Special is ±1.1 °C or ±0.4%. These apply only at manufacture and do not account for installation or aging. Use the tolerance calculator to find the limit for any type and temperature, and see the full tolerance class tables.
The In-Service Error Budget
| Error source | Typical magnitude | How to control it |
|---|---|---|
| Initial tolerance (type class) | ±1 to ±2°C class | Specify Special tolerance where needed |
| Cold-junction compensation | ±0.25 to ±1°C | Accurate CJC sensor; ice-point reference for precision |
| Drift / decalibration | Grows with time & temperature | Scheduled verification intervals |
| Immersion / stem conduction | Up to several °C | Immerse ≥ 20× sheath diameter; good thermowell contact |
| Extension-wire mismatch | Variable, can be large | Matched extension grade & correct polarity |
| Electrical noise / ground loops | Erratic | Ungrounded junction, shielding |
Where Accuracy Is Usually Lost
- Insufficient immersion. Heat conducted away along the stem pulls the junction toward ambient. Immerse at least 20× the sheath outer diameter into the medium.
- Cold-junction error. Because the reading is referenced to the terminal temperature, a 1 °C CJC error becomes a 1 °C measurement error. This is the most overlooked contributor.
- Wiring faults. Reversed polarity and unmatched extension wire can add errors larger than the sensor's entire tolerance — see troubleshooting.
- Drift. A thermocouple that was in tolerance when installed may not stay there; the inhomogeneity that develops in the temperature-gradient zone is why recalibrating a used sensor is often unreliable.
Combining the Errors
Independent error sources combine as the root-sum-of-squares (RSS), not by simple addition — the dominant term usually drives the total. In most industrial installations the largest terms are immersion error and drift, not the sensor's initial tolerance. Controlling installation and maintaining a verification schedule typically improves accuracy more than upgrading to a Special-tolerance sensor.