VoltRef - Voltage Scaling System The following simplified method compares the different outputs of voltage standards. It enables users to make comparisons between different outputs up to 50 volts. Comparisons can be made between 1, 1.018 and 10 volts to better than 0.5 ppm. This system is included in the VRMP software described in the previous section. The only equipment needed besides the software is a Low Thermal Scanner and a voltmeter with sufficient resolution and linearity. then taken as before. These readings are adjusted using the preliminary meter correction. Computed values are found for the 10-volt units from the least-squares-fit calculation that are biased on the known 1-volt values. The final meter correction is found by adjusting the correction so that the adjusted computed values for the 10-volt devices are equal to their adjusted measured values. By using the method described only the meter readings at 9 and 10 volts are significant in determining the final values for the 10-volt units. TEN-TO-ONE PROCEDURE The first step measures each traceable reference unit included in the test design. This set of measurements, along with a repeat set of measurements taken at the end of the procedure, is used to calibrate the voltmeter. Difference measurements are then taken between pairs of devices as described in the previous section except in this case the 9 volt difference between 1-volt and 10-volt outputs is measured. A calibration ratio for the meter is calculated by dividing the actual readings by the known value of each of the traceable reference units. This calibration factor is then applied to each of the difference measurements. The corrected measurements are used in the least-squares-fit calculation to determine the unknown values. When using 10-volt traceable reference units this procedure works very well. Since the readings between pairs will be approximately 9 volts, and the calibration is done at 10 volts, the voltmeter need have good linearity only over the range of 9 to 10 volts. The values for the 1-volt units are computed from the least-squares-fit calculation. ONE-TO-TEN PROCEDURE A more complicated procedure must be used to ensure good results when scaling from 1 or 1.018-volt units. Rather than simply calibrating the meter at 1 volt and depending on it to be linear from 1 to 9 volts, an alternative method is used that only requires the meter to have good linearity over the range from 9 to 10 volts. In this procedure the measured value for each 10-volt device in the test is taken along with the traceable 1 or 1.018-volt reference devices. A preliminary meter correction is found. The difference measurements are RESULTS The accuracy of the scaling procedure is primarily dependent on the meter linearity and on the resolution available on the 10-volt range. Tests using the procedure with two commercially available voltmeters have been done at laboratories that have 10-volt Josephson arrays. Many test runs show that values obtained with the scaling procedure lie within 0.5 ppm of measurements made directly against the Josephson arrays. Typical measurement differences were about 0.2 ppm. The two meters tested were the Hewlett-Packard 3458A and the Datron 1281. Several of each of these two models were used..Laboratories can now take advantage of the superior long-term stability of saturated cells to maintain their volt, while using the more rugged solid-state devices as working standards and for volt transfers between laboratories. Page 6