Metrology of Zener-based secondary voltage standards

Abstract

The BIPM (Bureau International des Poids et Mesures) has been carefully maintaining Zener diode secondary voltage standards (ZVS) for about 30 years, which are used in an ongoing international bilateral comparison program with national metrology institutes. The output voltages of the standards are corrected for their sensitivity to parameters such as their internal temperature (relative to the thermistor resistance of the Zener's oven) and atmospheric pressure. In 2016 and 2017, these sensitivity coefficients were redetermined. The corresponding corrections vary between −7 parts in 10 7 kΩ −1 to +6 parts in 10 7 kΩ −1 , and from +8 parts in 10 10 hPa −1 to +2 parts in 10 9 hPa −1 , respectively. In the worst case, the combination of the effect of temperature and pressure are comparable to the 1 f −1 noise floor of these standards (considered to be 1 part in 10 8 at the BIPM). Most of the sensitivity coefficients have not changed since their previous determination in 1998-2002 which seems to indicate that they are intrinsic characteristics of the ZVS, which are dependent upon the properties of the physical support of the Zener element itself (pressure sensitivity) and the intrinsic behaviour of the internal thermistor (temperature sensitivity). As a follow-up to the determination of the new sensitivity coefficients, the reproducibility of the ZVS calibration was also investigated, using three different and independent measurement setups, each built around a different and independent Josephson voltage standard. The metrological performance of the three primary setups were demonstrated to have an agreement of between 2 × 10 −10 and 1 × 10 −11 when compared directly. The comparison results of the calibration of the BIPM ZVS showed that most artefacts reach an agreement of a few parts in 10 9 while a very few can exhibit a considerable systematic error comparable to the 1 f −1 noise, which we suspect is due to a nonzero current from the measuring system flowing through the input impedance of the standard.

A 10 V PJVS-based DC voltage realization at INRiM

“…Error bars represent the standard deviation of the mean of five repeated measurements. However, it is widely recognized that the calibration uncertainty of commercial Zener-diode voltage references is dominated and limited by their inner flicker noise [10], which is likely to underestimate the actual random effects when the standard deviation of the mean is considered, especially for long measurement times. To account for the presence of the Zener flicker noise, a type-A uncertainty contribution not lower than 100 nV for the 10 V output is usually assumed [10].…”

“…However, it is widely recognized that the calibration uncertainty of commercial Zener-diode voltage references is dominated and limited by their inner flicker noise [10], which is likely to underestimate the actual random effects when the standard deviation of the mean is considered, especially for long measurement times. To account for the presence of the Zener flicker noise, a type-A uncertainty contribution not lower than 100 nV for the 10 V output is usually assumed [10]. With regards to type-B uncertainty contributions, these are associated to: i) null-detector gain and non-linearity; ii) uncompensated thermal offsets; iii) leakage resistances; iv) microwave frequency.…”