Calibration of the dual-mode auto-calibrating resistance thermometer with few-parts-per-million uncertainty

Abstract

The dual-mode auto-calibrating resistance thermometer (DART) has recently been proposed for highly accurate temperature measurement based on noise thermometry. In this paper, it is demonstrated that calibration and operation of the DART at part-per-million (ppm) level should be possible with the hardware developed. For this purpose, we have extensively tested a representative signal path comprising the basic DART components. This includes a low-noise amplifier connected to a 24-bit ADC and a metrology-grade voltage reference. A Josephson arbitrary waveform synthesizer (JAWS) generates a pseudo-noise consisting of low-distortion multitones superimposed on a low-frequency square-wave reference voltage. Using this signal, a fast and efficient calibration scheme for the signal path gain is demonstrated. The reference voltage stabilizes the gain at ppm level. We observed gain fluctuations within over a period of 19 d, a temperature coefficient of , and insignificant nonlinearity within an uncertainty band of for rms input levels between 5 µV and 80 µV. The behavior of the signal path with a 300 Ω resistor as a noise source was also investigated. From the observed stability of the voltage reference and flatness of the noise gain between 10 kHz and 225 kHz, we estimate that the presented hardware components are suitable for temperature measurements with systematic uncertainties well below .