American Institute of Electrical Engineers, Copyright 1909. By A. I. E. E. CONDITIONS AFFECTING STABILITY IN ELECTRIC LIGHTING CIRCUITS BY ELIHU THOMSON The term stability in this paper refers to that condition in which the current or potential, or both, remain at a steady value. Instability implies fluctuation which may arise from internal causes or be provoked by slight changes; and the fluctuations so arising may tend to be perpetuated automatically. An example of such action is the well-known "surging " in series arc-light circuits, which in the early days of development often became a difficulty. The chief part of the present discussion will naturally have to do with the constant-current series circuits containing arc lamps operated by direct-current dynamos, by alternating-current transformers, by reactive devices in alternatingcurrent circuits, or by rectified alternating currents from transformers. Much that will be touched upon is already well known; but there may be some matters which are not so well known, the bearing of which may be rendered more clear. The instability due to so-called "hunting" will be left out of present consideration, which will be confined chiefly to cases in which the load conditions themselves tend to instability. The art of electric distribution began about 30 years ago with the direct-current series arc circuit. The load was inherently unstable owing to the peculiarity of the arc resistance falling with increase of current. This form of distribution seems destined to survive for a long period, especially for street service with the later luminous flame arcs of high economy. It will also hold its place in similar service with series incandescent lamps with metallic or other high economy filaments, the nature of the resistance of which results in stability of current. Such a circuit as the latter is therefore outside of present consideration. Where in fact the load in a circuit is of the nature of "dead" resistance, stability is assured, provided no other disturbing causes exist, such as speed fluctuations in driving the generators. This is of course the case in the incandescent lamp circuit, whether in series, in parallel, or in a combination of the two. In early continuous current plants of small capacity, it was, however, generally easy to note the disturbances introduced by speed irregularities, or even to count the revolutions of the driving engine, or to watch the intermittent action of the engine governor. The coupling of machines in parallel, and the later introduction of turbine driving, have resulted in the almost complete elimination of such fluctuations and they may, therefore, be left out of consideration as having nothing to do with inherent lack of stability or with such instability inherent in the arrangement of the circuit itself. In passing, however, it may be remarked that sometimes in self-excited direct-current dynamos of constant-potential type, shunt or compound, the degree of sensitiveness to speed variations may be abnormal. This sensitiveness has occasionally been noted by designers in high-speed, high-efficiency machines of moderate capacity. In such cases each slight increment of speed raises the potential and immediately reacts to increase the field strength, in turn. causing a further increase, and so on. With a slight fall of speed there is of course the opposite effect. The difficulty is exaggerated if the engine governor is a little late in making its compensations, and they are at their worst when the iron of the machine is worked at the steep part of the curve of magnetization much below incipient saturation. The remedy for this form of instability is an alteration of proportions in the design and the removal of mechanical accessory causes. By working higher up on the saturation curve the tendency to magnification of the effects above alluded to is much reduced. The greater stability secured is, of course, purchased at the cost of more copper or a less efficient field circuit, and in a compounded machine the curve of potential regulation may not be so flat. It is fortunate that the great fall of resistance in carbon filaments occurs only at ranges of temperature much below normal incandescence, or such lamps might become a load tending to instability, as in the case of the Nernst glower, which requires the addition of devices for securing stability of current. The metallic filament and metallized filament, having positive. temperature coefficients, tend, if anything, to minimize all fluctuations, but the effect is naturally very small. Some tests made with rods of fairly pure silicon seem to show that, unlike carbon, it has a positive coefficient up to about 600° or 700° cent., beyond which the resistance falls very rapidly. Had, therefore, silicon possessed the refractory character of carbon, it would probably have been an unstable load, if otherwise suitable for filaments and tried in lamps. The Nernst lamp glower is like the arc in being an unstable load. An arc started between electrodes held at a fixed distance apart and maintained at a constant potential difference would exemplify the most unstable condition possible, but the requirement is manifestly impracticable. It would result in infinite current passing. The current would grow in value until the assumed condition of constant potential difference ceased to exist. The condition reached would virtually be that of short-circuit. While the conditions assumed above are unrealizable, they are sufficiently approximated in practice, and they emphasize the fact of the need of some current-limiting or steadying device in circuit with the arc. Several arcs in series evidently do not change the nature of the problem. By an inspection of Fig. 1, in which the curves represent the relation of current and resistance and of current and potential in an open arc with fixed separation, these curves being properly regarded as "characteristics," the unstable condition is manifest. An enclosed arc gives substantially the same result. It has been said with some truth that to design and construct a constant-potential dynamo that would work fairly well, was, even in the early years of the art, an easy undertaking compared with the production of a dynamo to operate a series of arc lights, involving a peculiar and little-known proportioning to secure stability of current in working a load so inherently unstable. The problem was not only this, but the stable current value had to be such as was proper to the machine, one which did not involve a destructive rise of temperature in the windings; and, on the other hand, one which was not so small in value as to involve such low temperatures in the windings as to indicate that the material was not being used at or near the maximum effectiveness. |