toward automation in business and industry, and the concomitant re-quirements of real-time computer operation, it is not surprising that the goal of a great deal of this effort is an all-electronic computer which achieves high-frequency response with no significant reduction in accuracy.
(1) The one component which is the subject of the greatest engineer-ing effort is the electronic multiplier. The time-division, diode, and AM-FM multipliers, discussed in Section 6.3, are results of this program. Although they are capable of high-frequency response (in the order of several hundred cycles per second), their accuracy (about 0.2 per cent) has not quite equaled that of the servo multiplier (between 0.02 per cent and 0.1 per cent). The complexity of these electronic multipliers is considerably greater than that of the servomechanism.
The search for a simple multiplier which obeys some physical law has produced some results but, unfortunately, these multipliers are not too accurate. can be synthesized by standard techniques.
(4) In Section 6.3 it was observed that recorders which possess high-frequency response are generally incapable of high accuracy, and conversely. It is often ironic that analog computer solutions stem-ming from mechanizations containing hundreds of components are gen-erally more accurate than are the means of recording them. The need exists for recorders whose accuracy and frequency response approach those of the computer itself.
(5) Associated with the problem above, as explained before, is the development of analog-to-digital converters for recording analog com-puter solutions accurately, although slowly. More generally, both analog-to-digital and digital-to-analog converters are the object of tremendous commercial activity directed toward application in the
176 AUTOMATION IN BUSINESS AND INDUSTRY
field of automation, in order to tie together analog and digital com-puters and instruments. This is the subject of Chapter 9.
(6) Several steps have been taken toward automating the in-sertion of a problem into an analog computer. Examples of this are the digital keyboard, used in conjunction with a servomechanism and clutches, for establishing the potentiometer settings, and the punch-tape input, also used to perform. this function as well as to establish the connections' between components.
(7) Some of the analog computers and simulators designed, more recently, for example the submarine simulator described in Section 6.1, include automatic check systems for performing static checks and, to a limited extent, dynamic checks. The static check permits the verification of all component connections other than the inputs to integrators; the dynamic check verifies the rates of integration.
6.7 REFERENCES
ANALOG COMPUTERS 177 16. E. A. Goldberg and G. W. Brown, An Electronic Simultaneous Equation Solver, J. Applied Phys., 19: 339-345, 1948.
17. J. H. Lanning, Jr., and R, H. Battin, An Application of Analogue Com-puters to Problems of Statistical Analysis Project Cyclone, Symposium II, Pt. 2, 79-87, Reeves Instrument Corp., N ew York, 1952.
18. R. R. Bennett, Analog Computing Applied to Noise Studies, P1'Oc. IRE, 1509-1513, Oct. 1953.
7. Digital ~ODlputers
WILLIS H. WARE
·7.1 INTRODUCTION
The RAND Corporation Los Angeles, California
The digital computer, or as it is becoming known, the data-processing or data-handling machine, is another tool for use in automation ap-plications. It is the second broad type of equipment which is avail-able for performing the computation necessary in many automation or control processes. Compared with analog computing, the digital tech-nique is relatively recent. Originally conceived for general engineering and scientific computation, machines constructed for this purpose were, in general, too expensive and too powerful mathematically for control or automation. It is only within the last few years that smaller, lower-priced machines have become available and have made economic the application of digital techniques to the automation field.
7.2 ANALOG VERSUS DIGITAL (1) 1
In the analog technique we obtain a solution to a problem by making measurements on a physical system. The simplest type of analog system is the model, in which a scale version of the problem of
in-1 Numbers refer to the references listed at the end of this chapter.
178
DIGITAL COMPUTERS 179
terest is constructed. From measurements made upon it the solution to the full-scale problem is achieved by extrapolating these measure-ments. Another example of the analog computing system is the di-rect analog, in whiah some alternate system is found whose equations of behavior are the same as the equations of behavior of the stated problem. Measurements made on the alternate system are then di-rectly transferable to the problem of interest. The last type of analog system is the differential analyzer, which may be either mechanical or electronic. In this case the machine actually solves the mathematical equations which describe the problem. In any type of analog system it follows that there is a direct correspondence between parts of the problem and parts of the computer. In the case of the differential analyzer, or electronic analog computer as it is sometimes known, the correspondence is between parts of the equations which describe the problem and parts of the computer.
In contrast, the digital system does not deal with a physical system per se, but instead it deals only with numbers representing the char-acteristics of the problem of interest. The principal characteristic of a digital computing system is that it counts or manipulates numerical expressions which represent physical behavior but which are dissociated from a physical system.
The problem of deciding whether a given system is digital or analog is sometimes difficult. There are fundamental characteristics of each type, however, and these indicate how a given system should be concerned with the counting or manipulation of numerical expressions.
In general it deals with information that is in discrete form; and it
Is this device counting something or is it measuring something?
Does this device deal with a physical· system or does it deal with numbers that are independent of a physical system?
The classical example of an analog computing device is the slide rule, which measures physical lengths, which in turn logarithmically