° ' ' ' ' ''.. ' ° "

D o n ' t th ink - c o m p u t e ! The advent of the electronic digital computer only 30 years ago and its subsequent proliferation, especially through the microprocessor, has resulted in a significant redirection of engineering design policy. Digital data processing has not only initiated a major change in design technique and philosophy, through the medium of Computer Aided Design (CAD), but it has also made possible a radically new approach to the control and operation of complete machines by putting a computer in the control system. This approach has made possible the simplification of the human operator's task (and in some cases to his virtual elimination) and also the develop- ment of operating modes and correction systems which would not have been possible ten years ago.

In some cases the decision to use sophisticated com- puter techniques may be questionable. The development of relatively cheap computer power during the last decade has encouraged some designers to abandon sound mechanical engineering principles (in both correctness of design and precision of manufacture) in favour of an apparently less expensive approach to the mechanics, associated with a plethora of sensors, actuators and computer algorithms which (it is hoped) wil l make due correction for any short- comings in the original design - nay, wil l actually produce (they believe) a more accurate machineT I do not believe that there is anything basically wrong in using a data processor to make a precision machine even more precise by applying on-line correction of systematic errors deter- mined previously during calibration, although the reaction of one precision engineer (J.B. Bryan of the Lawrence Livermore Laboratory) to this 'gadgeteering' approach to precision engineering was well stated in last October's editorial.

My reason for referring again to the subject is not to reply directly to Jim Bryan but rather to draw attention to the very serious lack of specialist systems-trained engineers capable of writ ing programs (software) for on-line real-time computer-based machines. Far too many organisations have short-sightedly invested in computer-based control and error-correction systems, persuaded no doubt by the temptingly low costs of sensors and electronic hardware, only to find that effective on-line real-time programming is very different from so-called business computing, and the majority of trained programmers currently available have only been taught the latter. Furthermore, a newly- hired programmer wil l not generally be familiar with a particular specialist field, eg diamond turning, and wil l accordingly be inefficient in designing computer programs for the specialist requirement; in some cases, the costs of writ ing and debugging on-line software have ultimately exceeded 20 times the hardware costs. The reason for this situation is not so much a shortage of programmers; rather a dearth of programmers with an in-depth knowledge and experience of the technology associated with the specialist field in which they are working. It is only too easy to appoint a 'programmer', and at the same time to be so naive to assume that this programmer wil l pick up in a few months the detailed specialist knowledge that some workers in the field do not ful ly acquire in a lifetime. The situation is not helped by the natural reluctance on the part of most engineers to take an interest in technologies outside their own narrow field of specialisation (this problem is well illustrated by the di f f icul ty in persuading mechanical engineers to work with electronics).

So what is the answer? One solution is to encourage some of your own engineers who are already experienced

in your specialist field (at least as far as conventional techniques are concerned) to take up computer programm- ing and learn to express their already proven flair in your field through the new medium of the computer. But, this does not mean tossing a copy of 'Teach Yourself Comput- ing'in their direction and suggesting they mug it up in their spare time. Neither does it mean sending them on a one- week intensive programming course at the local university or technical college. A number of companies have done just this, and subsequently proved to their bitter regret that real-time on-line programming is generally far more complex than business Programming. In fact, several organisations have become so disenchanted with data processing, as experienced in this way, that they have reverted to manual or semi-manual operation for their machines, and now go out of their way to discredit the computer in the ears of any who wil l listen

The inspired application of computer techniques to a particular specialist field, in such a way that all of the existing knowledge and techniques can be both applied and enhanced by the computer, is a highly-developed skill, part- science and part-intuition, that is not going to be acquired in a few months. What is needed is an on-going activity in computer applications, conducted in parallel with the design and development of machines using conventional tech- nology. The whole enterprise should be backed up with a regular schedule of informal lectures, (describing both conventional and computer-based approaches) given by members of the organisation and by specialists from out- side. In this way all members of the organisation wil l become more familiar with both technologies, an in-depth expertise in computer application engineering wil l be built-up, and hopefully resistance to the new technology wil l slowly soften as fear and suspicion give way to famili- arity and confidence. Until on-line real-time systems pro- gramming becomes a recognised discipline, and teaching methods have been perfected, organisations wishing to have at their disposal engineer-programmers who have a wide experience of their own specialist field and also a practical working knowledge of systems programming must also be prepared to release an appropriate employee to work for up to two years at a university or technical college with a research programme in systems programming. Organisations that have adopted these approaches to the introduction of new technologies frequently find that the decision to use new methods emerges from a consensus of opinion among the workforce, based on proven results on the shop floor or in the laboratory, as opposed to from a board-level directive made in ignorance, hope or desperation.

In spite of the remarkable advances made during the last two decades, the computer is still in its extreme infancy. With the exception of a few 'number-crunching' activities, the computer is of l itt le real use until it can be built into a complete system. The next 20 years will see a concerted effort to ensure that computers attain a healthy adolescence as components of machine systems in which the associated hardware and software interact with the rest of the machine elements to provide a higher performance at no great additional cost.

Maybe then we won't call them gadgets.

~ g J. Dinsdale, ' " n of Machine Systems,

Cranfield Institute of Technology, Cranfield, Bedford MK43 OA L, UK

P R E C I S I O N E N G I N E E R I N G 3