8.1      Accuracy considerations

          The original Dobson instrument electronic system (Dobson, 1931) was quite advanced for its day, though by today's standards it was relatively simple. It comprised a photocell, an AC amplifier, a motor driven chopper and commutator, and a microammeter display. Subsequent developments (Normand and Kay, 1952; Else et al., 1968) Olafson, 1968; Raeber, 1973; Komhyr and Grass, 1972; Chopra et al., 1977; Michiej and Sztanga, 1981; and Westbury et al., 1981) have altered various of the components of the system, and have thereby greatly improved its performance, but the basic design has remained unchanged. The replacement of the photocell by a photomultiplier was an especially important improvement. Komhyr and Grass (1972) describe a number of problems with the original componetry, in particular, bulky power supplies, tube amplifiers which were electrically noisy and for which parts were difficult to obtain, and dirty and electrically noisy copper split-disc electromechanical commutators. Most instruments nowadays are fitted with solid state amplifiers and power supplies and with optical switch "commutators", such as are described by Komhyr and Grass (1972).

          The electronics system has a number of features which prevent it from being an important error source. Firstly, the use of the null detection method means that there is no need for linearity of response or long term constancy of response. Secondly, the AC detection system is insensitive to constant DC sources, such as photomultiplier dark current. Thirdly, the 27.5 Hz (or 33 Hz) frequency of the chopper wheel/commutator device effectively rejects AC leakages, in particular that coming from 50 Hz (or 60 Hz) mains supplies.

          Component failures may make the instrument unuseable on occasion, but generally problems in the electronics system manifest themselves not as any systematic error, but as increased electrical noise which is seen as an increase in the unsteadiness of the microammeter needle. This results in an increase in the random error of the measurement and a loss of sensitivty, and hence in a reduction of the operating range of a given measurement type (e.g. AD direct sun). Komhyr and Grass (1972) and Komhyr (1980b) give practical details on the components, the adjustments and the maintenance required to obtain optimum performance.

          Photomultipliers can be a source of systematic error. Photomultipliers are valued for their great sensitivity, linearity and dynamic range, and the types used in the Dobson instrument are usually the RCA 1P28 or EMI 9781 side window types. (See manufacturers' handbooks, such as RCA, 1970; and EMI, 1970.) The instrument's relative spectral sensitivity constants Kk (see equations 1.6 and 1.7) depend directly on the photomultiplier's effective spectral sensitivity, and this can vary in several ways.

          Cathode spectral sensitivities are temperature dependent, but the dependences are poorly known. In the spectral region 300 to 350 nm the temperature dependence of caesium antimony cathodes is about 1.5x10-3% (°C nm)-1, (RCA, 1970). If this value is also typical of the RCA 1P28 caesium bismuth cathode and the EMI 9781 cathode, then the temperature dependence of relative sensitivity for a Dobson bandpair of 20 nm separation would be 0.03% °C-1, which would give, for a maximum temperature range of 30°C, a maximum range in the bandpair's extraterrestrial constant of about 0.004, or equivalently, a 1.5% range in the C bandpair's ozone estimate. The data for caesium antimony show the dependence to be similar for all bandpairs and hence its effect would be largely eliminated in double bandpair ozone estimate. Of course these are just estimates, and the actual temperature dependences remain to be determined.

          Cathode sensitivity and cathode spectral sensitivity can vary significantly spatially across a photomultiplier's cathode. Variations in effective spectral sensitivity therefore can occur if the spatial distribution of the illumination of the cathode is varied, and for this reason a small lens, L5 (see Figure 1.2), is placed at S5 to focus a fixed image of prism P2 onto a fixed area of the cathode. The photomultiplier's position must of course remain fixed. Also, the prism P2 must be fully and uniformly illuminated. Variations in effective relative spectral sensitivity can be detected, and hence corrected, by means of the standard lamp test, and such tests should always be done both before and after any adjustment or replacement of the photomultiplier or lens L5.

          Variation in effective spectral sensitivity can also occur as a result of variations in the cathode to first dynode voltage, owing to the combination of the spatial variation in cathode spectral sensitivity and the spatial dependence of cathode emission and dynode collection on applied voltage. A common method used to avoid this problem is to fix the voltage between the cathode and first dynode (and sometimes also between the first and second dynodes) by means of zener diodes, but this approach does not appear to have been taken with, or considered for, the Dobson instrument. The very great range in the ultraviolet light intensity being measured requires a very large variation in photomultiplier sensitivity and hence a large variation in overall applied voltage. Thus there certainly exists the potential for such spectral sensitivity variations.

          Photomultiplier sensitivity is very dependent on applied high voltage, by typically 10% for each 1% change in voltage, and photometric instruments generally need highly stable high voltage supplies. However, the Dobson instrument's AC null detection system is relatively insensitive to mains ripple, drift and other minor fluctuations, and so requires only modest stability from its photomultiplier power supply.

          It would be desirable for further investigations to be made of the temperature and voltage dependences of photomultiplier relative spectral sensitivity in Dobson instruments. The tests could be readily made with the aid of standard lamps of sufficient brightness.

8.2 Summary

(i)      The basic design of the electronic system is simple and robust, and the system is not an important source of systematic error under usual operating conditions.

(ii)     Deficiencies in the electronic system generally result in decreased sensitivity and increased signal noise and therefore in limited operational range and greater random error. The introduction of photomultipliers, solid state amplifiers and solid state commutating switches has significantly improved the system's performance and reliability.

(iii)    The photomultiplier's relative spectral sensitivity may have small temperature and voltage dependences involving ozone estimate errors of possibly more than 1%. These should be investigated further.

Return to Table of Contents

Forward to Ozone Absorption Coefficients