Abstract
A switching-type noise thermometer is used for the determination of the ratio of thermodynamic temperatures of two resistances, one having a value of 10 kΩ and maintained at a temperature in the range 90 to 100 K and the other, regarded as the reference, of 3 kΩ maintained slightly above room temperature (298 K).
The various instrumental errors are analysed and it is shown that operation with the instrument simultaneously balanced in two different frequency ranges results in the automatic elimination of some of these, allowing a precision of 1 in 105, equivalent to 1 mK at 100 K, to be approached, subject to the requirement of lengthy averaging for the reduction of statistical uncertainty.
Preliminary results, based on the assumption of negligible difference between the IPTS-68 and the thermodynamic scale at the reference temperature, show the difference T - T68 to be + 3 mK at 90.17 K and + 8 mK at 97 K. The standard deviations of these differences are 3 mK and 4 mK respectively.
Recent gas thermometry at NBS indicates that the IPTS-68 temperature may be significantly higher than the corresponding thermodynamic one at the reference temperature and if this is taken into account the noise thermometer result is in close numerical agreement with the IPTS-68 value for the oxygen boiling point.