Table of contents

When calibrating a sensor, the evaluation of the uncertainty of the estimated function is at least as important as the numeric values of the function itself. The calibration process usually involves a least-squares fitting, performed, for example for a thermocouple, by comparing some voltage readings with corresponding temperatures from a reference thermometer. The aim of this work is to encourage the use of improved algorithms, based on an extended minimization criterion, which can overcome some limitations of the classic method.

Procedures developed for accurate measurement of amount of substance ratios for gas isotopes, as used to improve our knowledge of the Avogadro constant and checked for quantitative conformity against kinetic gas theory, were used to measure synthetic mixtures of enriched xenon gas isotopes, prepared by (micro)-gravimetry as is usual in the preparation of synthetic gas compound mixtures. This allowed a totally independent evaluation, on the same samples, of the relative combined standard uncertainty of (a) the procedure to perform direct measurements of amount of substance ratios of gas isotopes in the redetermination of the Avogadro constant and (b) the gravimetric preparation procedure. The results of both procedures, mass spectrometry and gravimetry, agree to a relative uncertainty of 2 × 10^-3 for amount measurements of the major abundant isotopes. Thus it seems that a direct measurement of isotopic gas mixtures (e.g. of natural isotopic composition) is now possible for xenon isotopes- and probably also for other gaseous isotopes- without necessarily having to rely on ''calibration'' by means of values provided by measurements of gravimetrically prepared isotope mixtures. However, synthetic mixtures may be needed for validation and verification purposes, in particular for quality assurance (QA).

Geometrical constructions are proposed which minimize residual impedances of four terminal-pair (4 T-P) capacitance, resistance and inductance standards together with adaptors to other terminal systems. Some standards whose measured frequency dependence is small (up to 10 MHz) have been made using surface-mounting electronic capacitors. These standards facilitate the traceable calibration of 4 T-P impedance-measuring instruments. A 4 T-P to single-port adaptor having small, measurable residual impedances bridges the gap in calibration traceability between low and microwave frequencies. The resistance standards usually used for ac/dc transfer up to 1 MHz can also be converted to a 4 T-P definition with another adaptor.

Comparisons of the 1 V Josephson array voltage standard of the Bureau International des Poids et Mesures (BIPM) were made with that of the Istituto Elettrotecnico Nazionale Galileo Ferraris (IEN), Turin, Italy, in November 1996. The results agree within an overall uncertainty of less than 1 nV.

An optically pumped caesium-beam frequency standard (known as the KRISS-1) has been constructed at the Korea Research Institute of Standards and Science (KRISS) and the first measurements of long-term frequency stability have been made. Two methods were used to make the frequency standard immune to variations in the environmental temperature: an F = 4 → F' = 3 (σ + π)-polarized laser was used as a pumping source, and the difference between two sequential error signals was used as a feedback signal in the digital servo loop. A frequency stability of σ_y(τ) = 2.5 × 10^-14 at τ = 10⁵ s for a temperature variation of 4 °C was achieved with the KRISS-1.

An interlaboratory comparison of Josephson voltage standards has been made among fifteen national, industrial, and military standards laboratories in North America. The comparison was made at 10 V using a set of four travelling Zener reference standards. The uncertainty (2σ) of the comparison is estimated to be 17 parts in 10⁹. The measurements of all participants fall within this uncertainty, indicating that no significant difference has been detected among them.

The deviation functions for thermocouples, representing the difference between thermocouple emf and the corresponding reference function, are examined using data on 367 thermocouples of types R and S, submitted for calibration over the period 1984 to 1996. Calibrations were conducted with the CSIRO National Measurement Laboratory (NML) calibration facility, which thermoelectrically scans the wires, automatically collects comparison data to 1100 °C and incorporates data from the melting point of palladium. It is shown that deviation functions are well behaved, changing gradually with temperature, and that relatively accurate calibrations may be carried out with few calibration temperatures. For example, a calibration to ~1064 °C may rely on two fixed points, those of aluminium and gold, and a calibration to 1550 °C on three fixed points, those of Al, Au and Pd. Such an approach is feasible for Pt-based thermocouples because linear interpolation over temperature intervals of ~500 °C contributes no more than about 3 µV to the expanded uncertainty of calibration.

The volume magnetic susceptibility of moist air χ_m(ma) is needed to correct the results of certain experiments to vacuum conditions. This paper derives an appropriate equation for χ_m(ma) for which pressure, temperature, relative humidity, and carbon dioxide content are variable parameters. The derived equation also depends sensitively on the molar susceptibility of oxygen, for which the handbook values are inconsistent by about 1.4%. The equation resolves this discrepancy in part.