A Ratiometric Method for Johnson Noise Thermometry Using a Quantized Voltage Noise Source

Abstract

Johnson Noise Thermometry (JNT) involves the measurement of the statistical variance of a fluctuating voltage across a resistor in thermal equilibrium. Modern digital techniques make it now possible to perform many functions required for JNT in highly efficient and predictable ways. We describe the operational characteristics of a prototype JNT system which uses digital signal processing for filtering, real‐time spectral cross‐correlation for noise power measurement, and a digitally synthesized Quantized Voltage Noise Source (QVNS) as an AC voltage reference. The QVNS emulates noise with a constant spectral density that is stable, programmable, and calculable in terms of known parameters using digital synthesis techniques. Changes in analog gain are accounted for by alternating the inputs between the Johnson noise sensor and the QVNS. The Johnson noise power at a known temperature is first balanced with a synthesized noise power from the QVNS. The process is then repeated by balancing the noise power from the same resistor at an unknown temperature. When the two noise power ratios are combined, a thermodynamic temperature is derived using the ratio of the two QVNS spectral densities. We present preliminary results where the ratio between the gallium triple point and the water triple point is used to demonstrate the accuracy of the measurement system with a standard uncertainty of 0.04 %.