On a two-wire Pt100, every ohm of cable resistance shows up as about 2.6 °C of room temperature.
A Pt100 changes by about 0.385 Ω per °C, and with two wires the transmitter cannot tell cable from element. A 30 m run of 0.5 mm² copper is roughly 2 Ω there and back, which puts the reading over 5 °C high before anyone has looked at sensor class.
Three-wire connection compensates for the leads as long as the conductors match. Four-wire takes them out of the measurement. A head-mounted transmitter at the sensor solves it another way: the long run to the panel carries 4–20 mA, and cable resistance stops changing the reading.
With the leads dealt with, the element's tolerance class starts to matter. IEC 60751 gives class B as ±(0.30 + 0.005|t|) °C and class A as ±(0.15 + 0.002|t|) °C, which at 20 °C is ±0.40 °C against ±0.19 °C. In a store held at 15–25 °C the difference is academic. In a room specified at 20 ± 0.5 °C, class B alone accounts for most of the band.
The wiring also decides how the loop is checked. An element calibrated in a bath and refitted says nothing about the cable. Comparing the panel reading with a reference probe beside the sensor in the room does.
Humiley designs automation and monitoring to 21 CFR Part 11 and Annex 11.
How are the temperature sensors in your tightest room wired back to the panel?



