[Trade Journal]
Publication: American Institute Of Electrical Engineers
New York, NY, United States
p. 514-518, col. 1-2
Methods of Measuring the Insulation
of Telephone Lines at High Frequencies
BY E. I. GREEN (1)
Associate, A. I. E. E.
Synopsis—This paper outlines the problem of measuring the insulation of open-wire telephone circuits in the frequency range from 8000 to 50,000 cycles, and discusses a method which has been used in the experimental study of insulator losses at these frequencies. The paper includes a description of a special line which has been constructed for the testing of insulators, an explanation of the essentials of the measuring technique, and a brief summary of the results which have been obtained.
MEASUREMENTS involving the transmission of high-frequency currents over open-wire telephone lines began in the Bell System about 10 years ago, as a preliminary to the application of the first carrier telephone and telegraph systems.(2) Since that time, more or less continuous study has been given to the different problems involved in the transmission over line circuits of carrier frequencies ranging from about 3000 to 50,000 cycles. From the beginning it was apparent that the attenuation of open-wire line circuits at these higher frequencies is very much greater than in the voice range of frequencies, and undergoes wide variations due to changing weather conditions. Inasmuch as the attenuation is one of the most important factors in high-frequency transmission, its investigation has been very actively prosecuted along both theoretical and experimental lines.
The fundamental problem which originally presented itself was that of segregating the different losses which are experienced by the high-frequency energy trans¬mitted over an open-wire circuit. It was obvious that a substantial part of the increased attenuation at high frequencies resulted from the increase in wire resistance due to skin effect, but it was equally obvious that other sources of loss were also contributing in large measure to the attenuation. It was known that radiation was a negligible factor in the line losses. It was known also that the "leakage" of the insulators increased rapidly with frequency, and that there was no direct relation between the high-frequency leakage and the d-c. leakage, which had previously been used as the principal criterion of the condition of the insulation of circuits, but information regarding the precise nature of the leakage losses at high frequencies was lacking.
The theory underlying computations of the skin effect resistance of conductors was well established at that time, and the effective resistance of the wires could be readily determined. In order to study the other losses properly, however, it was necessary to develop methods and apparatus for more accurately measuring their magnitude. These methods and apparatus, and their application in practise, form the subject of this paper.
THEORY OF LEAKAGE MEASUREMENT
Transmission over wires at high frequencies is accomplished in precisely the same manner as transmission at low frequencies, the wires acting as the guiding medium for the energy in both cases, and the same fundamental equations may be applied to both. The rate of attenuation for a uniform line circuit which is terminated so as to avoid reflection effects may, therefore, be determined by means of familiar transmission formulas. The attenuation constant a determines the decrease in magnitude of the voltage and current transmitted over the circuit, according to the equations
where E1 and I1 are the voltage and current at the sending end of a section of the circuit, and E2 and I2 are the voltage and current at a point distant 1 units from the sending end.
The value of the attenuation constant at any frequency may be derived from the so called "primary constants" of the circuit, which are as follows:
R = series resistance in ohms per unit length,
L = series inductance in henrys per unit length,
C = shunt capacitance in farads per unit length,
G = equivalent shunt leakage conductance in mhos per unit length.
These constants determine the value of the well-known expression
of which the attenuation constant is the real part.
The symbol G in the above equation represents the "equivalent" leakage conductance. It is convenient to make this equivalent value of G include all of the a-c. losses suffered by the energy transmitted over the pair except the actual I2 R loss in the wires themselves.
An obvious method of finding the value of G under such conditions is that of measuring the attenuation of a section of line and computing G by means of the attenuation formula, using known values of R, L, and C. This method has been used quite extensively, and is of considerable value. Unfortunately, however, its use requires, for accurate results, a section of line of the order of at least 100 miles in length. Any important changes in a line of such length are quite expensive, and this method is consequently not well adapted to the experimental study of the equivalent leakage conductance obtained for different line arrangements and different conditions of insulation.
A much more satisfactory method for this purpose is to measure the leakage conductance on a short line. If the line is short enough to make propagation effects negligible, its impedance measured with the far end open will be
and the value of the leakage conductance may be obtained directly.
The determination of the maximum length of line for which propagation effects are negligible is comparatively simple. The problem is merely that of making the length of the test line a small enough fraction of a wavelength to produce only a small phase change in the current or voltage transmitted over the line. It can be shown that when the total phase shift on the line does not exceed five degrees it may be neglected without appreciable error. Now the phase shift for an open-wire pair is approximately two deg. per kc. per mi. Hence, if the phase shift which is allowed is not to exceed five deg., and the measurements are to cover the frequency range up to 50 kc., the maximum length of line which can be measured is evidently 0.05 mi., or about 250 ft. It has been found that the use of lines whose length does not exceed this value gives quite satisfactory results.
It should not escape attention that the use of a short line for measurements of the equivalent leakage conductance also involves the assumption that the short line provides all the sources of loss which are present on a longer line of, say, 100 mi. in length. The validity of this assumption has been tested by comparing the results obtained on a short line with those for a long line, and the comparison has shown that, whereas if the long line is infrequently transposed there may be some absorption of energy due to currents induced in the other circuits on the lead, the installation of transpositions required for minimizing high-frequency cross-talk ordinarily reduces such losses to a negligible value. This is only another way of saying that the shunt losses on a well transposed open-wire line occur almost entirely at the insulator points. This fact is one of outstanding importance, since it means that the insulation losses may be determined on a short line on which a considerable departure from the spacing of the insulators and the wires on a standard long line is permitted. The method of determining the leakage conductance on a short line is, therefore, extremely advantageous in the study of the effectiveness of different types of insulators at various frequencies and under various weather conditions.
INSULATOR TEST LINE
There is illustrated in Fig. 1 a short line which has been built near Phoenixville, Pennsylvania, for use in comparing the effectiveness of different types of insulators at high frequencies. This line includes about 25 poles spaced about seven ft. apart. A six-in. spacing between wires is used in order to make provision for the installation of a larger number of different types of insulators than could be obtained with standard wire spacing. About 40 different types of insulators are actually installed on the line in the picture.
In constructing this line, it was found convenient to make the length and spacing of the wires such that the value of the wire capacitance obtained was less than the minimum value desired for measuring purposes. With this arrangement, the capacitance may be increased to any desired value by shunting a condenser across the wires. The number of insulators installed on the wires was made slightly greater than the minimum value which was deemed essential for accurate measurements.
For a line only 175 ft. long, on which problems of external interference or interference between circuits do not exist, it might be supposed that the installation of transpositions would be useless. Upon investigation, however, it was found that an unbalanced relation between the two wires of a pair and the adjacent wires might produce appreciable loss at high frequencies. Accordingly, a simple transposition scheme for balancing the different capacitances between wires was devised and installed.
Owing to the comparatively small number of insulators employed on the line, it was necessary, in bringing the wires the test station, to use every possible precaution in order to avoid having the entrance losses comparable in magnitude with some of the insulator losses which it was desired to measure. This difficulty has been obviated by the use of several interesting expedients which are illustrated in Figs. 2 and 3.
Fig. 2 shows the pole at the entrance to the test station, while Fig. 3 shows the entrance arrangements at close range. Each wire is brought into the test station through a glass tube. A special arrangement of glass shields or baffles is built over the entrance, the three narrow panes used for this purpose being barely discernible in the picture. In order to prevent rain from running down the wire to the glass tube, each wire is equipped with a drip washer. Springs are used to keep the wires taut.
TESTING TECHNIQUE
For a test line short enough to avoid propagation effects, the impedance which is to be measured may be considered as a single conductance shunted by a given value of capacitance. This is equivalent, of course, to a leaky condenser, and a bridge method similar to those which have been employed for the determination of the loss angle or power factor of a condenser may be used in this case. Since the line wires have a large capacitance to ground, however, it is important that the bridge should be well balanced to ground.
The general arrangement of the bridge and associated apparatus used in a typical measurement of carrier frequency leakage conductance on the test line at Phoenixville is shown in schematic form in Fig. 4. An illustration of the physical disposition of the apparatus is found in Fig. 5.
The salient feature of this bridge arrangement is a specially designed transformer consisting of three air-core coils mounted in a shielded container. In accordance with telephone parlance, this three-winding transformer is ordinarily termed a "hybrid coil." The hybrid coil type of bridge is well adapted for use in measuring impedances whose center is at ground potential since the capacitances of the windings to ground can be balanced. The details of the design of the air-cored hybrid coil are illustrated in Fig. 6, the most important feature being the use of bifilar wire for the two windings to which the oscillator terminals are connected.
The source of high-frequency current for the bridge is a vacuum tube oscillator, and the detecting means is a detector-amplifier, both being of the types which are described in a parallel paper. (3) It is ordinarily assumed that the loss in the air condensers used in the measurement is negligible. In order to justify this assumption, considerable care must be exercised in the selection of the condensers.
In the frequency range under consideration, the equivalent resistance of a small number of insulators in parallel is quite high and is therefore difficult to duplicate on the standard side of the bridge. It is well known, however, that a condenser having some dissipation of ,energy may be considered, at a given frequency, as equivalent to a hypothetical resistance either in series or in parallel with a perfect condenser. Hence the line impedance, which resembles that of a condenser and resistance in parallel, may be simulated on the standard side of the bridge by a resistance and condenser in series. The range of values of standard resistance required with this arrangement can be physically realized without difficulty.
Instead of employing the obvious method of adjusting the capacitance on the standard side of the bridge to equal that on the line side, use is made of a method which is much more convenient for the computation of results. With this method the capacitance on the line side of the bridge is adjusted to the capacitance on the standard side. Thus condenser C1 is kept at a fixed value, and the capacitance balance is obtained by varying the setting of condenser C2 .which is shunted across the line. The resistance balance is obtained by adjusting RI. Only the frequency and the value of R1 need be recorded for any measurement.
The theory by which the recorded values of R1 and frequency in combination with the known value of CI may be converted into the desired value of leakage conductance is outlined in the Appendix, where it is shown that the unknown leakage conductance G2 is given by:
This equation shows why it is unnecessary to read the value of condenser C2 used or the unknown side of the bridge. It also indicates the dependence of the value of the balancing resistance R1 upon the total capacitance on the unknown side of the bridge.
The balancing of the bridge is a rather delicate matter and extreme care is required in order to secure accurate results. The reason for this will be apparent when it is noted that the reactance represented by C1 may be, in dry weather, several hundred times the value of the resistance R1. The difficulty of attaining a high degree of accuracy in the reading of R, when the voltage across it is only a few thousandths of that applied to the bridge, scarcely needs to be pointed out.
In setting up the apparatus, every precaution is taken to avoid any stray pick-up of the oscillator output in the bridge or detector circuits. Such a mischance is ordinarily obviated by keeping the various pieces of apparatus well separated.
The results of a typical set of measurements of leakage conductance at different frequencies are plotted in Fig. 7. As indicated in the figure, it is ordinarily desirable to record in some detail the weather conditions which prevail at the time when the measurements are made. In the correlation of the measured values of leakage conductance with the weather variations, a recording rain gage has been found very useful for indicating the total precipitation and the rate of precipitation throughout any testing period.
The record of the a-c leakage conductance is generally supplemented by a record of the d-c. leakage under the same conditions. This latter may be readily obtained with a source of direct voltage and a micro-ammeter or a high-resistance voltmeter. It has also been found desirable to secure continuous records of the d-c. leakage over fairly long periods of time, and for this purpose a recording micro-ammeter has been used with satisfactory results. By means of a multi-record instrument, simultaneous records may be obtained on as many as six pairs. The general circuit arrangement for a continuous measurement of d-c. leakage is indicated schematically in Fig. 8, and a sample record for a single pair is given in Fig. 9. The rainfall record corresponding to the leakage measurements of Fig. 9 is presented in Fig. 10.
It has been recognized for some time that a method of obtaining a continuous record of the a-c. leakage conductance on one or more pairs would be extremely valuable, since it would provide a record throughout the night, when the test station is normally closed and might economize on testing time during the day. The problem of developing such a method has been attacked from several angles, but no completely satisfactory result has as yet been attained.
CONCLUSION
In conclusion it may be said that the general methods described above have been used at Phoenixville, Pennsylvania, in the study of the performance of various types of insulators over the entire frequency range up to 50,000 cycles. The work has served to illuminate the different phenomena involved in the leakage conductance of open-wire lines, and has made possible an accurate determination of the relative magnitudes of these phenomena. Finally, it has resulted in the development of insulators which have improved characteristics in the carrier range of frequencies and which are now rendering service on many lines of the Bell System.
The author wishes to state that this paper describes work in which a number of engineers of the Department of Development and Research of the American Telephone and Telegraph Company have been engaged. Particular credit is due to Mr. R. N. Hunter, who developed the air-cored hybrid coil, and began the measurements of leakage conductance, to Mr. L. T. Wilson, whose researches into the theory of insulator losses and their measurement have been of great value, and to Mr. F. A. Leibe, who has made important contributions to the measuring technique.
1. American Tel. and Tel. Co., 195 Broadway, New York, N. Y.
2. See Carrier-Current Telephony and Telegraphy, E. H. Colpitts and 0. B. Black-well, TRANSACTIONS A. I. E. E., Vol. XL, pp. 205-300,1921.
3. High-Frequency Measurements of Communication Lines, H. A. Affel and J. T. O'Leary, TRANS., A. I. E. E., 1927. p. 504.
Presented at the Regional Meeting of District No. 1 of the A. I. E. E., Pittsfield, Mass., May 25-27, 1927.
