|Digital Hardware Design
I. CATT D. S. WALTON M. DAVIDSON
CATT, D. S. WALTON, M. DAVIDSON
The authors, who have long experience of research and development in digital electronics, became concerned at the lack of attention paid to the design of digital hardware by educational establishments. This book is based on material developed for a series of seminars which are now attended by engineers drawn from the design laboratories of many large electronics companies in the United Kingdom.
If and when colleges and faculties of electronics and physics realise that digital hardware design is a major discipline, this and subsequent volumes planned by the authors will be invaluable for the design of relevant and up-to-date courses.
All engineering is based on science, as academic establishments quite rightly emphasise. However, it is essential that the science which is taught is relevant to current practice and technology in engineering. If this is not so, the student comes to industry armed with a set of scientific principles, and corresponding mathematical techniques, which are inappropriate to his work as a designer and which lead to disillusionment with science in general. Worse still, the diligent application of inappropriate techniques may lead to bad design, often with disastrous consequences.
The digital electronics industry has sprung up so quickly in the last 10 years that the theoretical foundation required has not developed at all. It is impossible to cross the line separating the analog and digital worlds. The sine wave is a periodic time-varying steady state situation, whereas a digital signal is a fixed amplitude step (shock wave). Each change of state is a single event in time and cannot be correlated with any other change. A dubious connection, via Fourier analysis, is merely a mathematical arpeggio, guaranteed to be worth a few exam questions at least. A leading edge of a step is a shock wave; it is a transverse electromagnetic wavefront which travels at the speed of light. Of course, it is possible to take this single step and analyse it using Fourier analysis but this would mean combining an infinite number of sine waves which exist from minus infinity to plus infinity. This can be easily seen to be quite absurd and of no practical use.
At the turn of the century Oliver Heaviside and his contemporaries Lodge, S. P. Thompson and Hertz developed many theories which should be used today. By thinking of digital signals as small discrete packets of `energy current' flowing at the speed of light between the wires (which merely act as a guide) many of the present-day design implementation problems could be solved. The advent of the telephone and radio led to the predominance of sinusoidal time-varying signals, so the concept of `energy current' was lost as new theories were developed to cope only with the periodic waveform. We have now turned a full circle and must look backwards before we can advance. The works of Oliver Heaviside contain a huge amount of relevant information and any serious minded digital engineer would gain enormously by reading his books.
The practical problems related to digital systems such as
need to be studied. General models and original concepts based on Heaviside's `energy current' idea can be used to tackle the above problem areas making it possible to design complex digital systems in an orderly scientific fashion. Every practising engineer in digital electronics must stop attempting to use analog ideas for digital systems; they will not work. This is easy to see; all around the industry are scattered systems which `crash' regularly. Pattern sensitivity, noise, power supply problems are all raising their ugly heads, and all quite unnecessarily. By following clearly defined design rules, systems can be built which will work reliably and first time, without the usual 3 to 6 month commissioning troubles. It is difficult to assess the financial saving that could be made if digital systems were developed using adequate theoretical principles. Suffice it to say that the saving would be significant. Also the job satisfaction of development teams would increase.
The hard and fast rules laid down for periodic sine wave situations must be cast aside and new rules developed for the shock wave situation. An obvious area to concentrate on is the one of signal distribution. Any prime source of electrical energy, be it analog or digital, needs to be easily distributed to loads that require it. We must have a basic understanding of the mechanism by which a block or pulse of energy is transmitted in space. This leads us into the realms of electromagnetic field theory, for it is here that the student will learn and ultimately understand the subject of digital electronics.
Unfortunately, nearly all the books written on the subject of electromagnetic field theory are concerned with steady state sine wave situations. There is no basic theory written today which concentrates on high speed digital techniques. The knowledge of how 1 ns steps propagate is known by only a few people Yet with the advent of ECL (emitter coupled logic) and Schottky TTL this electrical phenomenon is becoming widespread. Engineers today
attempt to put together fast, complex logic systems which are doomed to failure. The paper design might well be satisfactory but the problems that arise during testing and commissioning seem endless. The unfortunate engineer just cannot understand the `gremlins' that keep upsetting his system. This is because nowhere is he taught the important fundamental principles necessary for competent digital system development.
In order to have a complete understanding of high speed systems one must apply certain techniques which are not taught in any educational establishment, nor written about in any textbook. One must go back to the turn of the century to find any suitable material. Then, the main subject area was telegraph signalling which is analogous to digital transmission today. A 10 ms risetime step or edge travelling 1000 km (telegraphy) is based on the same theoretical principles as a 1 ns step travelling 10 cm (computers).
Finally, and probably the most important point, not one of the design concepts that are used is difficult. Although soundly based in theory, they do not involve exotic mathematics and are aimed specifically at practical problems of hardware development. They are tools of the trade to be used by all engineers and technicians. There is no need to allow ourselves to be surrounded by a fog of complex but inappropriate mathematics, when there is the chance to gain a clear understanding of a challenging, high technology industry.