"About a volt" isn't good enough: standards based on quantum-mechanical phenomena have performance unmatched by man-made artifacts

EE-Evaluation Engineering, May, 2006 by Tom Lecklider

Urban Legend 37: Speeding case thrown out of court because police radar gun had not been calibrated for five years. This defense argument might not work for you although radar-derived evidence has been found unreliable in several well-publicized speeding cases. (1)

For example, "In [New Jersey] State v. Wojtkowiak, 174 N.J. Super, 460 (App. Div. 1980), the appeals court held in all future cases the state should adduce evidence ... as to (1) the specific training and extent of experience of the officer operating the radar, (2) the calibration of the machine was checked by at least two external tuning forks both singly and in combination, and (3) the calibration of the speedometer of the patrol car in cases where the [radar gun] is operating in the moving mode." (2)

Conclusions in similar cases question the accuracy of the calibrating tuning forks and refer to the need for certification of calibration documentation for police radar equipment. Of course, these kinds of concerns have long been recognized by the test and measurement community. Anyone who has worked in a calibration-conscious organization is familiar with the colored labels glued to test equipment stating the last and next calibration dates.

[FIGURE 1 OMITTED]

A dictionary defines calibration as the act of checking or adjusting, by comparison with a standard, the accuracy of a measuring instrument. For example, police-car speedometers can be calibrated to read the true speed within 1% rather than the 0% to 10% faster reading typical of a passenger car.

Whatever measurement capability is being calibrated, a standard of known accuracy is required. If an instrument's measurement readings agree with standard values to within the instrument's specified accuracy, no adjustment is required. The instrument already is calibrated, and documentation only will confirm its adequate performance. However, if the instrument readings are outside the specified error margin, then adjustment is required. A detailed certificate of calibration may list before and after values but should always state the final values to prove that the specified accuracy has been achieved.

Voltage Standards

For many years, the most accurate source of a known voltage was the Weston standard cell (Figure 1). Although these devices were commonly found in standards labs throughout the world, they were finicky and delicate at best. The output voltage was nominally 1.01830 V at 20[degrees]C and varied slightly with temperature and atmospheric pressure in well-understood ways.

A group of cells was maintained by a lab and each cell periodically measured by comparing it to one of its peers. The value of the reference voltage the lab used in calibration work was based on a statistical combination of the voltages from each of the cells in the group.

Connecting two cells in series opposition meant that just the voltage difference was being measured. Typically, cells differ by no more than a few microvolts, so even using measuring equipment with only kilohm-level input impedance ensured that current was at the nanoamp level. Ideally, no current should be drawn from a standard cell to avoid disturbing its voltage. Modern instruments based on semiconductor technology achieve greater than 100-M[ohm] input impedance, making routine cell comparison straightforward. The Weston standard cell is a type of artifact standard.

The standard volt now is defined in terms of a Josephson junction oscillator, a so-called intrinsic standard. The oscillation frequency of a Josephson junction is given by

[f.sub.Josephson] = [2e[DELTA]V]/h

where the relationship between frequency and voltage across the junction depends only upon the fundamental constants e, the charge of an electron, and h, Planck's constant. For 1 [micro]V applied to the junction, the frequency is

[f.sub.Josephson] = 483.6 MHz

The standard volt now is defined as the voltage required to produce a frequency of 483,597.9 GHz. (3)

Because voltage and frequency are linearly related, practical primary voltage standards are based on an array of thousands of Josephson junctions running at a lower frequency (Figure 2). The Josephson junction consists of an insulating barrier between two superconductors. The exact composition of the superconductor isn't important, but the entire array must be operated at 4[degrees]K to meet the superconducting requirement.

Josephson junction arrays represent the most stable and predictable standard voltage source. In comparisons between similar array-based standards, differences in the nanovolt range were recorded. Just as labs could buy Weston cells, voltage standards based on Josephson junction arrays are commercially available.

Sandia National Laboratories has both a laboratory and a portable array and quotes uncertainty on the 10-V range as [ or -]0.017 ppm. This compares with about 0.11 ppm for a group of Weston cells.

A 1985 IEEE paper described an automated DC measuring system that the National Bureau of Standards, now the National Institute of Standards and Technology (NIST), had developed. The author commented, "Typically, Kelvin-Varley dividers are used to measure a wide range of voltages in terms of a standard cell. Accuracies better than 1 ppm are only achieved by frequent, time-consuming, manual calibrations of the divider."