Software in scientific instruments-a good practice guide

Almost all of the current generation of scientific instruments contain a significant amount of software. Since it is hard to quantify the reliability or quality of such software, two questions immediately arise. (1) As a user of such an instrument, how can I be assured that the software is of a sufficient standard to justify its use? (2) As a developer of such software, what development techniques should I use, and how can I assure my customers of the quality of the resulting software? This good practice guide addresses these two questions. The guide surveys current good practice in software engineering and relates this practice to applications involving scientific instruments. Known pitfalls are illustrated with suggested means of avoiding them. The guide also covers a number of standards which specify requirements for software in scientific instruments. The general approach is a three-stage process: (1) a risk assessment based upon a model of an instrument with its software; (2) an assessment of integrity required on the software, based upon the risk assessment (called the software integrity level); and (3) guidance on the software engineering methods to be employed, determined by the software integrity level. There is no simple universal method for producing correct software and therefore skill, sound technical judgement and care are required. Moreover, if it is essential for the quality of the software to be demonstrated to a third party, then convincing evidence is needed which should be planned as an integral part of the software development process. To aid in the application of this guide, some check lists are provided