摘要:

Investigations into the characteristics of spark discharges have been made for impulse currents of up to 100 kA peak. The generation of such currents is based on the techniques developed for surge-testing purposes to simulate lightning-stroke conditions. For oscillatory impulses the current is determined primarily by the discharge energy of the generator and the total effective inductance. The generator described in the paper has been constructed to have a low inductance (0.1 μH) and gives a peak current of 470 kA for a discharge energy of 18000 joules at 25 kV. It is being used for studies of the spark channel under different conditions.The measurement of impulse currents is normally made with the aid of a low-resistance shunt. A shunt of coaxial form, giving a small time-constant and freedom from unwanted coupling with the main current circuit, has been used with the above generator and is described.

DOI:10.1049/pi-4.1954.0001

出版商:IEE    

年代:1954

数据来源: IET

摘要:

It has been shown that, if a plane wave is incident normally on a flat conducting ribbon, both components of induced current density differ little from what would obtain if the width of the ribbon was infinite, save very near the edges. The density very near the edges was evaluated for ribbons whose total width was 2λ/π or 4λ/π it emerged that the distribution of density was very nearly the same in each case and did not differ significantly from the distribution at the edge of a very wide ribbon, which is known from another solution. In other words, the distribution very near an edge is sensibly independent of the width of the ribbon.An example of great practical interest is the current induced in a flat reflecting sheet by a current filament parallel to it, or—speaking generally—when the incident wave has a cylindrical, in contrast to a plane, front. The exact solution for this case is known in principle but is impracticable to evaluate numerically. However, the solution can be evaluated when the ribbon is very wide indeed and the cylindrical wave starts from a line in the vicinity of the single bounding edge. Experience indicates that it seems reasonable to suppose that the “edge effect” for this limiting case will not differ significantly from the effect at both edges when the cylindrical source is situated on the normal to a ribbon of finite width.The distribution of induced current has been evaluated when the cylindrical source is (a) distant ¼λ from a half-plane and 0.592λ from its bounding edge, (b) distant 0.96λ from a half-plane and 0.555λ from its bounding edge, (c) distant ½λ from a half-plane and 0.99λ from its bounding edge. In each case the disturbance of density is found to be concentrated in a width of about 0.1λ from the bounding edge.Then follows a Section which shows that the net electric force at the foot of the perpendicular from a filament distant ½λ from a plane is almost zero when the induced currents which are more distant than 0.99λ from this point are ignored. And that the small residue of force is sensibly equilibrated by the force which arises from the effect of two bounding edges, as estimated from the edge effect which obtains in a corresponding half-plane. The Section is intended to show that the calculable edge effects appropriate to a half-plane provide a very close solution for a ribbon of finite width.Then it is shown that the diffraction pattern for a filament in the presence of an infinite conducting plane depends dominantly on the currents which are induced in the width, say, ±λ from the foot of the perpendicular.Curves from which the diffraction pattern can readily be deduced are given for a filament distant from a plane and due to the currents induced in the central strips of various widths; they are intended to be of practical use as an aid to design.The modifications to these patterns which would result from the estimated “edge effects” which must occur when the reflector has a finite width, in contrast to the pattern accruing from the current induced in a finite width of an infinite plane, are considered and found to be insignificant.Accordingly, it is suggested that the diffraction pattern for a filament and a reflector of finite width can be estimated very closely by crediting the reflector with the currents which would be induced in the same width of an infinite plane. It is held that this has established a principle of considerable use in practice.Although an aerial and a flat reflector is not a case of much practical concern, at least it is one which can now be worked out simply and with considerable accuracy; it appears that there is little to be gained from making the reflector wider than 3/2λ.The principle developed in the paper must be capable of wide extension, and it should be possible to extend it to apply to cylindrical and to parabolic reflectors. It is hoped to do this in due course.

DOI:10.1049/pi-4.1954.0002

出版商:IEE    

年代:1954

数据来源: IET

摘要:

A new method of determining the external dynamic characteristics of an electric arc, without reference to physical events occurring within the arc, is described, and a testing procedure is developed. The principles are introduced in the first place by reference to a simple example, and a summary of more general rules is given in an Appendix. Although the method can be fully described without reference to any others, a brief account of the arc equations of Cassie and Mayr is given to illustrate essential differences.There follows a description of the first attempt to apply the theory, in which tests were carried out on a 1-cm carbon arc carrying currents of the order of 20 amp. Part of a first-order differential equation is obtained and comparison is made with the Cassie and Mayr equations. A consistent form of discontinuity was observed during the testing and a criterion for this is given.This method of approaching the problem raises some questions about electrical conductors or circuits in general which may eventually prove to be important. The discussion of these is brief, as it is necessarily speculative.

DOI:10.1049/pi-4.1954.0003

出版商:IEE    

年代:1954

数据来源: IET

摘要:

Methods are now well established for analysing the behaviour of closed-sequence control systems in which the relations between the quantities may be represented by differential equations, the inputs being known functions of time.The paper is concerned with the methods of analysis when these conditions are not met, the system having definite time delays or random disturbances or a combination of both. This occurs, in particular, in all manually operated controls, because there is an unavoidable time delay of relatively large duration in all human muscular response, and the behaviour of even a skilled operator is never completely consistent. The variability of response can usually be considered to be equivalent to an additional input of a random nature, corresponding to the operator's inability correctly to assess the situation and to apply the signals corresponding to his assessment. Analogous difficulties arise in process controls and in many servo systems.It is shown how the behaviour of a system with a definite time delay and a random disturbance may be calculated and represented in non-dimensional form for cases in which the remaining part of the transfer function has certain simple but rather general forms. Results are given and compared for two alternative methods, namely the solution of the differential equations and the solution of finite-difference equations which represent approximately the same relations. It is shown that a finite-difference representation using a time interval equal to the finite time delay gives tolerably good results and requires much less labour.Charts are given showing the extent of amplification of the random disturbance.More specifically, the paper presents, in convenient non-dimensional form, a complete picture of the performance of any control system having an open-loop transfer function of the form (1+T1p)(1+T2p)ε−Tp/P2with a random disturbance. This includes, as special cases, manual controls either of the rate, position or aided-lay varieties, certain types of process control, and, as a first approximation, a considerable range of servo-mechanism systems, since the definite delay term ε−Tpmay be considered as an approximation to the effect of a number of small time delays in cascade.A further form of loop transfer function to which some consideration is given is [(1+T1p)(1+T2p)ε−Tp]/[(1+T3p)(1+T4p)P2], which applies to an even wider range of servo mechanisms.

DOI:10.1049/pi-4.1954.0004

出版商:IEE    

年代:1954

数据来源: IET

摘要:

The paper considers the restrictions which must be placed on the poles and zeros of a system function of a linear system with lumped constants in order that the response of the system to a step-function drive should be monotonic. Necessary and sufficient restrictions are derived for a system function (driving-function/response-function) containing up to three zeros, and these are interpreted as permissible pole-zero positions on the complex plane for the one- and two-zero cases and also for a number of instances in the three-zero case.The paper concludes with a method of obtaining sufficient conditions for monotonic response in the case of more complicated system functions from the results given in the paper.

DOI:10.1049/pi-4.1954.0005

出版商:IEE    

年代:1954

数据来源: IET

摘要:

The magnitude and distribution of cable irregularities are related statistically to the end-to-end input-impedance difference under conditions involving a large number of wavelengths and appreciable attenuation. The manner of variation of the measured quantities with frequency and irregularity magnitude and distribution is revealed. These relationships are shown in the form of master curves for a particular cable on the assumption of exponential fault correlation.

DOI:10.1049/pi-4.1954.0006

出版商:IEE    

年代:1954

数据来源: IET

摘要:

The paper describes a method of square-wave (on-off) modulation of the reflector electrode of a reflex klystron which produces a stabilized frequency during the “on” time. During the “off” time the voltage applied to the reflector electrode is outside a mode of oscillation and the klystron does not oscillate. The modulation is applied through a special blocking modulator circuit which ensures that the duration of the frequency sweep immediately preceding stabilization is kept to a small value. The stabilization is accomplished by means of a frequency discriminator unit, which feeds back a voltage through the blocking modulator (while the klystron is oscillating) thus stabilizing its frequency. It is therefore essential to obtain a smooth voltage/frequency discriminator characteristic over the swept frequency ranges to eliminate the possibility of stabilizing at more than one frequency. Both theoretical and experimental characteristics of the microwave discriminator are discussed. It is shown that the output-voltage/frequency characteristic of the discriminator is critically dependent on the phasing of sidebands and the (frequency sensitive) phasing of a carrier at a detector crystal. It is also shown that, measured in the plane of the detector crystal, the phase of the carrier should be in quadrature with that of the reference phase of the sidebands, not only at the resonant frequency of the controlling cavity resonator, but also at frequencies well removed from resonance.The theoretical analysis would be useful in other situations where the relative amplitudes and phases of a carrier and two sidebands have to be considered.

DOI:10.1049/pi-4.1954.0007

出版商:IEE    

年代:1954

数据来源: IET

摘要:

In the insertion-loss design method for filters, compensating networks and other frequency-selective 4-terminal devices, the final step is the synthesis of a network of coils and capacitors which possesses a set of characteristic functions of frequency, in the form of one of several alternative kinds of matrix, previously deduced from the desired overall frequency behaviour. The present paper is chiefly concerned with this step.Practical experience has shown that the most satisfactory way of synthesizing a complicated network is to use a chain of sub-networks connected in cascade. For such a synthesis the most convenient matrix is the so-called “chain matrix,” since the chain matrix of a chain of networks is simply the product of the individual chain matrices. Conversely, a specified chain matrix can be synthesized by factorizing it into sub-matrices, each of sufficiently small degree to allow of direct synthesis, and then connecting all the corresponding networks in cascade.In previous methods of factorizing a given chain matrix it has always been necessary to split off small network sections one at a time. In the present method sections of arbitrary size can be split off in each stage. The importance of this is that the larger the size of section to be directly synthesized, the greater is the likelihood of being able to avoid coupled coils in the design.The method depends on a new theorem of a novel and far-reaching character concerning reactance and impedance functions. In proving and discussing this, reference has to be made to certain properties of such functions which are of great importance in themselves but have not hitherto been presented in a connected manner. Accordingly, a systematic treatment of these is first given, including a number of results believed to be new.A possible extension of the method, potentially of considerable value, to general dissipative networks is indicated.

DOI:10.1049/pi-4.1954.0008

出版商:IEE    

年代:1954

数据来源: IET

摘要:

The paper presents a theory of capacitor bushings for both infinite and finite numbers of foils. Two bushing contours are considered, namely the straight-line contour and the contour which gives uniform axial potential gradient. The general line followed is to determine for each contour (a) the radial stress in the insulating material, (b) the location of the place of minimum radial stress, (c) the flange radius, and (d) the minimum flange radius.Consideration is given to the case where the conductor radius is fixed, and also to the important practical case where the length of the earth flange is fixed.The theory of capacitor bushings has been well covered by previous publications for the case of an infinite number of foils. The present paper extends this previous work, particularly in respect of the method of locating the position and magnitude of the minimum radial stress. The published theory of capacitor bushings is meagre for the case of a finite number of foils. The present paper covers new ground in the following: (a) the theory of a bushing with straight-line contour; (b) the location and magnitude of the minimum radial stress; (c) the theory when the limiting stress is that occurring at the last foil but one, instead of at the last foil.Throughout the paper the method of determining minimum flange radius differs substantially from that given previously. Since for each contour the cases of infinite and finite numbers of foils are considered, this permits of a determination of the error made if results obtained for an infinite number of foils are used when the number of foils is finite.The paper does not include the bushing theory for uniform radial stress since this has been dealt with in several earlier publications.

DOI:10.1049/pi-4.1954.0009

出版商:IEE    

年代:1954

数据来源: IET

摘要:

Synchronization of a certain type of valve oscillator, having a low-pass filter in the synchronization loop, and also of self-starting synchronous motors is examined. Under some conditions the pulling effect takes place in these circuits, i.e. the pull-in range of synchronization is smaller than the pull-out range. The pulling function, i.e. the relation between parameters of the circuit and the ratio of the pull-in and the pull-out range, has previously been examined only qualitatively in relation to synchronous motors. The quantitative examination presented here yields simple formulae applicable to design problems. An asymptotic approximation is also found to the pulling function in a synchronized system possessing a limitation in the synchronization loop

DOI:10.1049/pi-4.1954.0010

出版商:IEE    

年代:1954

数据来源: IET