The spectrometer frequently does not stand alone but is part of a system, the spectrometer installation.

First there is the item to be analysed. The analysis can control the production process up to the final quality. This happen eg. in metal manufacture in the form of charge monitoring. Samples are regularly taken from the furnace and master alloys, ores or scrap are added on the basis of the analytical results until the required alloy is obtained.

      Calculation for the type and quantity of additions are known as charge correction calculation and these are carried out today with computers dedicated to spectrometers.

      Sometimes part of the sample preparation process is carried out at the sampling point , but the sample is frequently transported direct to the laboratory. After preparation it is conveyed to the spectrometer the radiation is geometrically dispersed. The intensities of required on capacitors. The measurement values can be processed in different ways. As regards the type of further processing, both technical and economic considerations are important. Because of the reduced cost of electronic modules and computers, spectrometers equipped for data processing are state of the art. It is advisable to carry out non-spectrometer-specific calculations and evaluations not on the spectrometer computer but, after data transfer , on a separate computer.

       The analysis must go back to the source and initiate reactions. Fully automated production control by spectrochemical analysis is not yet a reality , due to three gaps which have only been partly closed :

       a) The first to be mentioned is automatic sampling. Progress has been made here and more will follow. Examples are：

       - sample-taking for effluent analyses,

       - sample-taking from crude ore blends for sinter production or crude cement powder, from the conveyor belt,

       -sample-taking from liquid metal baths.

        b)Samples are being increasingly delivered to the radiation source by automatic means. Previously, XRF spectrometer were more suitable for sample delivery automation because of the clear process cycles in the “stand “, compared with AE spectrometers. This situation has altered because of a better understanding of the process with SDAR.

        c)The third gap in the chain leading fully automatic process control is difficult to close, namely automation of the reaction on the basis of analytical results. In many cases this requires radical intervention in the production process, at considerable expense in terms of equipment and organization. The computer-controlled addition of master alloys for metal refining after charge correction calculation is being considered.

        The efficiency of AE spectrometer is due to the following features:

         a)  speed,

         b)  simultaneous detection of numerous analytes in one cycle

         c)  simultaneous detection with LOD in µg/g for most elements, and simultaneous determination of medium and high concentration with sufficient accuracy

         d) relatively low cost per analysis compared with other methods.

         Some supplementary explanations:

         a) Simultaneous AE spectrometers with SDAR have cycle times of 10-20s, including measurement evaluation and printing-out of 20 elements. The time required to run through the system is, on the other hand, lengthy. With appropriate automation and organization, a time of 3 min is achieved from sample-taking to complete feedback of analysis results.

         b) About 70 elements can be determined. These include the commercially important metals and semi-metals on which the widespread use of AE spectrometer is based. Some gases can be measured by means of special facilities. Representativity and homogeneity of sample must be carefully watched.

         c) There is a widespread view that AES with SDAR is not suitable for high concentrations, This has led to the use of XRF spectrometer. This view is only partly justified eg. when determined the composition of ferro-alloys or base elements in ores or alloys. In these cases, the accuracy of AES with SDAR is not good enough. The GDL gives good accuracy for high concentrations. As the calibration curves are largely linear, correspondingly good SR are obtained. SDAR with the HEPS technique also gives a good RSD of precision for intensity ratios of 0.1-0.5% for concentrations of 1-40% for metals. As the calibration lines are mostly curved, the RSD and SR are poorer. This problem can be solved, however, by appropriate analyte and homologous reference lines and of course with discharge parameters which lead to linear calibration.

        Element concentrations are determined, as measurements are carried out with radiation from atoms(not molecules). If eg. Ca is present in slag samples as sulphate, carbonate, oxide or in another form, the total Ca concentration is determined, not various compounds.

         d) Spectrochemical analysis of 20 analytes in steel costs as much as for one analyte in the chemical laboratory. To give this indication, the cost in the chemical laboratory was divided by means of automatic combustion equipment for comparison. Particularly efficient are spectrometer installations for a frequently recurring analysis programme. Then costs per analysis rapidly decrease with increasing number of samples per month.

         Notes on quality control of spectrometers for routine analyses:

         After the individual components have been assembled in functioning order the spectrometer goes to Quality Control , where mechanical, electrical, optical and analytical tests are carried out according to a Quality Control schedule and documented in test records.

         The aim of Quality Control is to confirm that the required quality has been produced during manufacturing and assembly, and only in exceptional cases to undertake rectification of any deficiencies.

         Unfortunately, fulfilment of the tasks described above is still only a rosy ideal for spectrometer manufactures’ Quality Control departments, although spectrometers have been built for more than half a century. Quality Control work still mainly consists of production-finishing.

         The reason for this must be sought in the fact that production and assembly workers are not sufficiently informed about what they are doing and why they are doing it. Thus things are often done in a purely mechanical way, with the motto “More grease spreads better”, and thus end up with the opposite of what should be achieved in a special case . This problem can only be alleviated by training, which can most simply be provided in rotation ie. People get to know what the “man in front” and the “man behind” do in the production line. Quality Control workers must spend some time in the assembly and calibration departments; assembly workers in components production and Quality Control; and calibration workers should also spend time in Quality Control. This promotes knowledge, motivation and thus quality.