Thermal EMF

Whenever two different metals join under a temperature gradient, the Seebeck effect produces a small DC voltage proportional to the temperature difference. Inside a real resistor this happens at the boundary between the resistance element (e.g. nichrome or Manganin wire) and the terminations (typically tin-plated copper or solder). The constant of proportionality is the Seebeck coefficient — about 1 μV/°C for Cu-NiCr and as low as 0.05 μV/°C for the carefully chosen Cu-Manganin or Cu-Karma pairs used in precision shunts.

For power resistors a few microvolts is negligible, but in precision DC contexts it dominates: a 4-wire 100 mΩ current shunt measuring 1 A produces only 100 mV signal — a 2 μV/°C × 5 °C gradient = 10 μV is 100 ppm of full scale, which destroys 0.1 % accuracy. Resistance bridges, voltage references, ovenized standards and 7½ digit DMMs all worry about thermal EMF.

Designers minimise thermal EMF by (1) symmetric layout so both Kelvin sense leads see the same temperature, (2) low-EMF terminations such as Cu/Cu-Manganin or copper-clad-copper solder joints, (3) airflow / shielding to suppress gradients, (4) AC excitation that ignores DC offsets entirely (used in many bridges and chopper amplifiers). Precision shunt datasheets quote a thermal-EMF figure in μV/°C; below 0.5 μV/°C is the bar for 6½-digit instrumentation.