VoltRef  Voltage Comparison Software 

The Voltage Reference Maintenance Program (VoltRef) completely automates the process for maintaining statistical control of the laboratory volt. It controls the scanner and voltmeter to take measurements, computes statistical information, prints reports, and stores the results in a data file. A graphics routine is included to create statistical control charts from the stored data. The system can make measurement runs on groups as large as 32 units. Also, the system can be easily set up to run measurements at night, avoiding disturbances. Special tentoone and onetoten scaling routines are included to allow comparisons between 1, 1.018 and 10volt standards. Scaling accuracies of 0.2 ppm can be realized using certain 8½ digit multimeters. These procedures are described in the next section. 
The program creates an appropriate test design that compares each of the units with other selected units. Comparisons are made for pairs of standards. For each comparison the voltmeter takes ten readings in the forward direction. Relays are then reversed and ten additional reading are taken in the reverse direction. This process effectively reduces errors caused by zero offset in the meter circuit. When all intercomparisons are complete, the program computes a leastsquaresfit of the data. Values for each unit are assigned based on this calculation and the values of the reference units included in the test. The procedures used in the VRMP are accepted by most organizations running Measurements Assurance Programs (MAPs) such as NIST and Fluke. In this way laboratories undertaking MAPs need only send in the printouts provided by the software. 

SYSTEM DESCRIPTION This system uses the procedure of measuring the differences between pairs of voltage standards in a measurement design. The program was designed to follow the procedure recommended by NIST to measure pairs of devices in a statistically balanced design. However, other methods such as the favored cell and ring designs can also be used. The scanner is a dual scanner that allows any two devices connected to the inputs to be compared against each other. All other devices connected to the scanner are left floating to reduce errors due to ground loops. The software provides a measurement design for any number of devices between 3 and 32. A delayed start feature will allow up to four different designs to be run unattended at any time. A scanner setup screen in the program makes it easy to change the arrangement and to add or remove units from the system. Once a unit’s name and channel number are entered, the program keeps track of its connection and switches it automatically in the test design. Values for the reference units are stored on a setup screen and are the values traceable to your national laboratory reference unit. 
DATA PROOF VOLTAGE REFERENCE REPORT STANDARD DEVIATION: 0.08
DATE: 8 July 1995 EMF DIFFERENCE BETWEEN PAIRS  MICROVOLTS
LEAST SQUARES ANALYSIS OF EMF STANDARD DATA GROUP NOMINAL MICROVOLTS GROUP MEAN 10REF 10 000 000.00 10 000 033.87 SCANNER UNIT
REFERENCE DEVIATION CORRECTED 



VOLTAGE SYSTEM RESULTS Standard deviations for a measurement design are typically less than 20 nanovolts at 1 volt, and 200 nanovolts for 10 volt units. This corresponds to 0.02 ppm. Thus, the laboratory volt can be improved through automation that renders the measuring system errors negligible. STATISTICAL
ANALYSIS 

A plotting routine that creates several different control charts is included in the software. The figure above shows the voltage for a 10volt standard over the period of one year. The vertical axis is 10 microvolts full scale or 1 ppm of 10 volts. The plot shows that the standard is drifting downward about 0.5 ppm per year with a fairly constant slope. The software can also plot the data with the slope removed. The figure at the left shows the same standard without the slope. Note that the vertical scale has been reduced to 0.4 ppm of full scale. If slope values for the reference units are entered on the setup screen, corrected values will be computed for each test. 


Home Page  Contents (Page1)  Previous Section  Next Section 