Practical Noise Thermometry Optimized for Speed and Measurement Uncertainty

Today, dc SQUID-based Johnson noise thermometers (JNTs) are widely accepted and applied as practical thermometers in millikelvin cryostats. Commonly known types are the Current Sensing Noise Thermometer and the Magnetic Field Fluctuation Thermometer. While the underlying theory is well understood, an aspect in their application that has been often overlooked is their intrinsic (non-thermal) noise. Its determination can be difficult and a few calculations are still missing it. We present a nonparametric temperature estimation model applicable to all JNTs that considers intrinsic noise, too. This comprehensive model of the JNT determines relative measurement uncertainty depending on temperature, evaluation bandwidth, averaging time and intrinsic noise. We suggest for temperature estimation that these quantities are dynamically adapted by an optimization depending on the actual temperature and application scenario of the JNT.