Temperature Coefficients of Direct Voltage Reference Standards at National Research Council Canada

Abstract

We experimentally determine the temperature coefficients of direct voltage reference standards known as Fluke model 732B at the National Research Council Canada. Two units, used as references, are maintained at ambient temperature for the entire measurement duration of 4 weeks. Ambient conditions are monitored using calibrated environmental data loggers. The standards under test are placed into a temperature-regulated air bath, where calibrated environmental data loggers monitor bath conditions near the voltage terminals and near the thermistor terminals. The following quantities are measured: voltage difference between standard under test and reference standard for both 10 V and 1.018 V channels, thermistor resistance of all standards, environmental conditions both inside and outside the air bath. The air bath temperature is set to stay on alternate weeks at higher and lower temperatures following the sequence (25, 20, 25, 20) °C. The ambient temperature is between 21 and 22 °C. The entire process is repeated a second time with the locations of the standards under tests and reference standards inverted. Data analysis consists of calculating the average temperature-induced voltage and resistance changes and dividing the results by the measured temperature change to obtain the temperature coefficients. A detailed uncertainty analysis is performed. The results are compared to manufacturer specifications. The majority of our standards are better than specifications, even when taking into account the measurement uncertainties. Such experiments are beneficial, as they allow the identification of the standards with the smallest temperature coefficients to be used as direct voltage reference in client calibration services. The results can be combined into a thermistor resistance coefficient, which can be used to provide a quantitative estimate for the size of the largest temperature-induced change of resistance that has negligible effect on the voltage outputs for a given tolerance level.