US761995A - Apparatus for reducing attenuation of electrical waves. - Google Patents

Description

No. 761,995. A PATENTED JUNE 7,' 1904. A M. I, PUrI-N.

- APPARATUS FOR REDUCING ATTENUATION OF ELECTRICAL WAVES.

AAAAAA ATION'IILBD TEE. 6.190s.

Inventor, .Mcha'e] I. Pu in,

v Affum gy'.

ITO-761,995- I Y O Patented June 7, 1904; UNITED STATES PATENT OFFICE.

MICHAEL I. PUPIN, OF NEW YORK, N. Y., ASSIGNOR TO AMERICAN BELL TELEPHONE OOMPANICOF BOSTON, MASSACHUSETTS, A-CORPORATION OF MASSACHUSETTS. V

APPARATUS FO REDUCING ATYTENUATIAON OF ELECTRICAL wAi/Es.

SPECIFICATION i'ormingpart (if Letters Patent No. 761,995, dated June 7, 1904. Application filed February 6, 1903. Serial No. 142,193. (No model.)

To all lull/0711' m yfi .drawings indicate. Fig. 4 represents a twin Be it known that I,MICHAEL I. PUPIN, aciticoil, each coil ot'the pair consisting of four 5 zen ofthe United States of America, andaresiparts, which are connected in series and ardcntof NewYorkcity,'borough of Manhattan, ranged symmetrically around the magnetic in the county and State of New York, have incircuit, Fig. 5 represents a twin coil in which vented certain new and useful Improvements each coil ofthe pair consists of four parts con- 4 in Apparatus for Reducing Attenuation of nectedin parallel and arranged symmetrically Electrical Waves, of which the following is a around the magnetic circuit. l specification. I have shown that theoretically the lamina- I O This invention relates to improvements of 'tion in a core forming a closed magnetic cirinductance-coils suitable for loading wave-doncuit can be sufficiently refined and the number ductors. The loading should be in accordance of ampere-turns can be suificientlyreduced to with the directions given in my United States obtain negligibly small iron-core loss of en- Patents Nos. 652,230 and 652,231, dated June ergy; but on account of its cost in many 'prac-. 5 19, 1900. It is known that such inductancetical cases an inductance-coil constructed in coils can be constructed with or without iron this manner will not answer, although from a cores. O'oils without iron cores occupy too theoretical point of view it is an ideal type of 5 large a volume for a given inductance to be coil. I find from astiidy of the practical consuitable for loading telephone-cables. Iron ditions in telephonic'and telegraphic trans- 20 cores, on the other hand, introduce internal mission that coils with cores forminga closed losses of energy which increase rapidly with magnetic circuit can be constructed which will 1 the frequency, and thus produce distortion of fulfil all practicalconditions and at the same 7 the transmitted complex harmonic W2LV8S-+ time be eificient'and, economical. These p rac such, for example, as accompany speech transtical conditions will now be discussed in their 5 mission. ltis known that iron-corelosses'ean proper order. Then an inductance-coil with v be diminished by a suYtIiciently-high degree of an iron-core forming a closed magnetic cirlamination and a suitable diminution of the cuit will be described which satisfies all these ampere-turns of thewinding. In the case of practical conditions. Finally, a method will cores which do not form a closed magnetic be disclosed which will enable one skilled in 3 circuit this is easily accomplished. It, ho'wthe art to construct an efficient and econom' ever, the core forms a closed magnetic cirical coil of this kind. The inductance-coils cuit, then a much stronger magnetization reconstructed in accordance with this method sults from a given number of ampere-turns, and wave-conductors loaded with such coils and accordingly a much larger core loss. In form the invention disclosed herein.

35 order to reduce for a given inductance the IIz'rst 0072.rZ'it7f"02?,-.A coil suitable for loadcore loss in this case to a'predetermined limit, ing a wave-conductor should for a given initis necessary to employ a much finer Subdiductance;,be .as small as possible. A coil of 5 vision of the iron and a smaller number of agivenz-inductance will occupyasmall volume turns. if its winding is distributed over an iron 4 In the accompanying sheet of drawings,v core ,forming a closed magnetic circuit, as which forma part of this specification, Figshown in Figs. 1 and 2 of the drawings and ure 1 is an end vie w,-and Fig, 2 is a side view, in the specification of my United States patof a twin coil having a core made up of plates ents above referredto. A still greater reducof iron. Fig. 3 represents a long-wave contion in the volume is obtained by placing two 4-5 ductor having a numberof twin coils, such as windings, one for the outgoing and the other are illustrated in Figs. 1 and 2, inserted at for the return wire of the circuit, onthe same periodically-recurring points along the concore. In this manner the inductance of each ductor, preferably at equal distances, as the winding is increased by the mutual inductance ductor, because such a disturbance would produce external induction, which in the case of telephone-cables would result in cross talk.

The symmetrical distribution of a wave with respect to a symmetrical wave-circuit is not disturbed when for each coil in the outgoing wire there is an equal coil at a homologous point of the return-wire. This distribution of'the coils preserves the symmetry of the wave-conductor.

\Vhen the windings of the two sides of the wave-circuit are placed on the same iron cores, they should be equal and stand in the same relation to the core. In other words, the two windings must have the same inductance and resistance. Their capacity should be negligibly small and can easily be made so, since the thickness of the insulation commonly employed is sufliciently great to insure this condition.

Tied rd condition.There should be no mutual inductance between inductance-coils belonging to different circuits when placed near each other, as in the case of loading a telephone-cable carrying a large number of circuits. This is. accomplished by distributing the winding of a coil over the iron core in such a way that no appreciable external magnetic field is produced by the transmitted waves. Such a distribution is a solenoidal distribution. The most convenient form of a solenoidal winding is a uniformly-distributed winding over a ring-shaped or toroidal iron core.

There are conditions under which it is indispensable to employ coils with iron cores. For instance, in loading underground and submarine cables economy in space demands efficient coils of large inductance per unit volume. In such cases iron should be employed. Economy in space is not the only consideration which recommends the employment of iron cores. Two other very important considerations will be stated here. "hey are: First, for all frequencies which are of importance in telephony the resistance corresponding to a given inductance can be made considerably smaller in the case of a coil with a properly-eonstructed iron core than in one having no iron core; second, it is much more easy to construct a practical form of coil possessing no external induction when an iron core is employed than without it all that is required is that the core should be uniform throughout and of symmetrical form and that the coil should be distributed over it symmetrically and uniformly.

I shall now briefly describe various forms of solenoidal windings. In the case of twin coils referred to above-that is, coils consisting of two equal windings on the same core each winding can be uniformly distributed over the ring core, so that one winding covers entirelv the other and each winding by itself will be solenoidal. This arrangement is the closest approach to theideal solenoidal winding, but not quite as economical as some other solenoidal forms; but in this case it would be somewhat diflicult to give each coil the same imluetanee and thesame resistance. To obviate this difficulty, the following arrangement is recommended: Wind two coils of thesame uuniberol' turns each of which is uniformly distributed over the uniform iron core. l,hey will completely overlap and should be properly insulated from each other. Divide each coil into two equal parts and connect each inside coil to its diametrically opposite outside coil. \Ye. shall thus obtain two equal coils each of which is uniformly distributed over the uniform core.

Figs. 1 and 2 of the accompanying d rawings represent an arrangement which under certain conditions will possess the properties of a solenoidal winding. A ring-shaped iron core f is built up of iron plates of predetermined thickness. Iron wireeanbe used for this purpose just as well, butnot so economicaltv as regards space. Two equal windings M and /1 are connected to the two sides of a circuit on this core. Each winding is distributed over one-half of the ring core. The two windings together form a solenoidal winding, and they will therefore produce no external field when properly connected in a symmetrical circuit. Other forms of windings to produce. the same result are represented in Figs. 4 and 5 of the accompanying drawings. in Fig. e1; each winding is divided into four equal parts connected in series, and the various parts of each winding are distributed symmetrically over the ring core. The arrangement represented in Fig. 5 differs from that of Fig. 4- in the manner in which the various parts of each Winding are connected. .In the arrangement of Fig. 5 the connection is a parallel connection. In order to obtain the same inductance and resistance in the parallel arrangement, it is necessary to have in each part four times as many turns and sixteen times as much resistance as in the series arrangement. In this arrangement (re n'esented in Figs. -1 and each winding by itself is approximately solenoidal, and this approximation is the closer the larger the number of parts into which each winding is divided. W'hen there are 91 parts to each winding, then for a given inductance there must be 1/ times as many turns and w" as high resistance in each part of theparallel arrangement as in the-series arrangement. It is easily seen that the weight of copper and the space occupied by it are very nearly the same in each case. The parallel arrangement would be preferable inthe case of coils-of high inductance and small ohmic resistance, such asare required in the loading of overhead telephonic long-distance wave-conductors of low resistance. In this case large iron ring cores are required and a heavy copper wire for the winding. A sub: division of such copper winding must be introduced in order to avoid eddy-current losses. These losses are serious, particularly for higher frequencies which occur in telephony, unless thediameter of the copper wire is suitably reduced. It proportional to the cube of the'diameter of the wire, asIh'ave demonstrated mathematically and verified experimentally. A very safe rule to follow is this: Do not wind coils containing iron cores for loading telephone and telegraph wires with wire of a greater diameter than .09 cm. Copper wire of diameter equal to .05 cm. gives exceedingly good results. In ballast-coils containing no iron core the subdivision of the copper wire should be considerably higher-say one-half of the size given above particularly when a low-resistance wave-conductor is to be loaded. This subdivision can be most conveniently effected in cases of this kind by using line Wire in the various parts of the two windings and then connecting these parts in par.- allel in accordance with the rule just given. The parallel arrangement requires more turns than the series arrangement; but where the 1s not of coils are wound by a machine this great moment.

in Fig. 3, is such that thecurrents which atany moment circulate in the two windings should magnetize the core in the same sense;

otherwise the inductance of one winding will not be increased by the mutual" inductance with the other Winding. To illustrate this point, suppose that we replace the coils ab 0d of Fig. 3 by the coils represented in Fig. 1 and Fig. 2, then the terminals 9 it 71 I. will fall upon the terminals a b 0 (Z, respectively. If we replace it by the coils represented in Fig. 4, then the terminals Z on a 0 will fall upon the terminals (0 6 0 (Z, respectively. Finally, if we replace it by the coil represented in Fig.

5 then the terminals 7) Q r s will fall upon the terminals a b 0 (Z, respectively. 1

Fowrzf/i condition. This condition refers to the lamination of the iron core and of the I copper winding in order to reduce energy losses due to eddy-currents. It is a fact that in coils the construction of which satisfies the three conditions named above the energy losses due to eddy-currents in the winding can be easily made very much smaller than those 1n theiron core, Wl11Cl1 latter w1ll therefore receive the principal attention here.

- for each apparatus be previously determined.

These formulae all refer to apparatus working with high magnetizations, such as are usually employed in ordinary commercial work. They have no application to theconstruction of inductance-coils intended for loading wave-conductors considered in this application where extremely-weak magnetizations come into operation. Besides, the method of construction of iron cores for inductance-coils which forms the foundation of this invention takes into account not the absolute value ofthe iron-core loss of energy, but the exact increase in-resistance of thecoilwinding due to this loss. Since the increase in cfliciency of transmission resulting from loading a wave-conductor depends on the amount of resistance introduced per unit of inductance by the inserted inductance-coils, it is evident that increase of resistance due to iron-core losses is a Very important quantity. There is no basis whatever in the existing state of the art for exact calculations relating to this subject.

The question as to how much resistance given inductance-coils with iron cores will introduce into a properly-loaded wave-conductor cannot be answered broadly, not even in the case of the coils of so simple a construction as is considered here. I have found, however, by my own investigations and experiments that within limits of magnetization which by suitable precautions can be made to satisfy all practical requirements in preperly-loaded wave-conductors for telephony and telegraphy this question can be answered with mathematical accuracy. 7

On pages six and seven of my United States patents above mentioned 1 have described an inductance-coil in which the increase in resistance due to iron-core losses is small per unit ITO Such inductance coils are perfect from a purely technical point of view, becausea properly-loaded wave-conductor say a loaded telephone-cable-will have with these ideal inductance-coils the same attenuation constant for all frequencies which it is intended to transmit. Hence its transmission will be distortionless. As above stated, however, from an economic point of view these ideal inductance-coils are in many practical cases not acceptable on account of the expense which their construction involves.

lt is a well-known fact that in telephony particularly and to a certain extent in telegraphy also a certain amount of distortion is necessarily introduced by the terminal apparatus without materially interfering with the quality of transmission. It is evident, therefore, that the resistance of the loading-coils due to iron-core losses need not be negligibly small, provided that the distortion caused by it is a small part of the distortion already ex isting in consequence of the terminal apparatus. The constructor of the inductance-coil must be guided, therefore, by the permissible distortion on the one hand and on the other hand by economic considerations which fix the cost of the coils within definite limits. It is evident, therefore, that between the extremes of an ideal coil of prohibitive cost and a cheap coil which introduces considerable distortion there must exist an interval of ironcore losses within which the coil will satisfy both the conditions of economy and of efficient distortionless transmission. This interval for telephonic transmission has been determined by me experimentally, and it can be described as follows: The increase of resistance due to energy losses in the iron cores of the loading coils should for a frequency as high as 1,500 p. p. s. be smaller than the true ohmic resistance of the copper of the loaded wave-conductor and greater than one-tenth part of such true ohmic resistance. It is easily seen that in loading a high-resistance telephone-cable the lower-resistance limit will be approached, and in loading a low-resistance air-line an approach to the upper-resistance limit will be sufficient. The true ohmic resistance of the loaded wave-conductor includes the true ohmic resistance of the inductance-coil windings, which can always in practice be much smaller than the true ohmic resistance of the wave-conductor before loading and is, in fact, usually about one-fourth in the case of low resistance air-lines and one-tenth in the case of high-resistance cables. It will therefore be seen that the rule that the increase in the resistance due to energy losses in the iron cores of the loading-coils shall be smaller than the true ohmic resistance of the copper of the loaded wave-conductor means, approximately speaking, that it shall not be greater in any coil than five times the true olnnic resistance of the winding of the coil, and

the rule that the resistance due to energy losses in the iron cores shall be greater than one-tenth part of such true ohmic resistance means in the case of coils for loading cables that the increase of resistance due to iron-core losses shall be greater than the true ohmic resistance of the coil. lherefore, expressing the rule so as to refer to the coils themselves without reference to the loaded comluctor, the rule of economical construction herein set forth requires that the increase of resistance due to iron-core losses of the coil shall for a frequency of 1,500 p. p. s. and for current strengths used in telegraph y and telephony be not less than the true ohmic resistance of the coil nor more than five times such true ohmic resistance.

The problem of loading telegraph-wires is much simpler than the problem of loading telephone-wires, because the frequencies involved are as a rule considerably lower than those involved in telephony. In considering the construction of coils for telephony l have taken a frequency of 1,500 p. p. s. as the stand ard. For telegraphy the standard frequency will be generally considerably lower, and the rules given here will be applied to the stand ard frequencies selected. These physical and economic considerations are necessary, because in order to satisfy the first, second, and third conditions mentioned above it is necessary to employ coils with iron cores forming practically closed magnetic circuits over which the winding is symmetrically distributed. In such coils the iron-core losses of energy corresponding to a given inductance and magnetizing force are much greater than in coils with open iron cores. To reduce these, it is necessary to laminate and employ a much smaller number of turns in the windings. Hence the excessive cost of construction if the increase in resistance due to iron-core losses of energy is to be negligibly small. .1 have established by theoretical and experimental lllV0.Htl, I:\tl()HS two mathematical laws which enable one skilled in the art to construct ind uctamze-eoils, giving for any frequency and current strength for telephonic or telegraphic purposes a predetermined increase in resistance per unit of inductance and will therefore enable the eonstruetor to build coils which will fall within the limits specified. These investigations will be published in the near future. The resulting laws will be found to be in harmony with everything disclosed in my prior patents and papers relating to loaded conductors. These two laws will be explained now.

1 72's! luwu ln Figs. 1 and 2 of the accompanying drawing let the iron core be coinposed of ring-shaped iron plates, which are insulated from each other. Let/1 equal external diameter of the plates, 0 equal internal diameter of-the plates, (1 equal thickness of one plate, 11 equal number of plates, equal permeability of the iron, equal specific resistance of the plates in C. G. S. units, f equal frequency of, a simple harmonic magnetizing force, equal number of turns in the winding, L equal inductance of the coil, R1 equal resistance of the coil due to Foucault-current losses in the core. Then we shall have the following relations:

- sistance telephone-cable (170 ohms per mile} 'of the plates will be (Z equals .0075 cm.

proper adjustment of the number of turns as specified furtherbelow we can have equals 100, at least, and (7 equals 1.3x 10 in C. G. S. units at most.

As a specific example let L equal .2 Henrys,

and let it be required that R1 equals 8 ohms.

at a frequency of 1,500 p. p. s. The thickness coils placed every half mile in a high reof circuit) or every two miles in a low-resistance telephone air-line (at ohms per mile of circuit) will satisfy all the requirements of correct loading, and the increase in resistance due to Foucault currents will be within the interval specified aboveth at is, provided the hysteresis loss can be kept down to certain small values.

The thickness of plates indicated is well within economic limits.

The construction of the inductance-coils considered here must satisfy the low hysteresis loss condition in order to keep the in crease in resistance, due to iron-core losses down to a predetermined limit. .In order. to

" avoid too high hysteresis losses and yet obtain the inductance given above, the size and number of plates and the number offturns in the winding must be properly determined.

It is a generally accepted belief that, according to Lord Rayleighs investigations, there is no hysteresis loss in hard-drawn iron so long as the magnetizing force per centimeter is less than .04 in C. G. S. units. Under these conditions there would be then no increase in resistance in the magnetizing-coil due to hysteresis. I find, however, that this supposition is wrong and can lead to results which are very detrimental to the efiiciency of the inductance-coils here considered.

Second Zm0.-The loss is small and cannot be detected by Lord Rayleighs methods; but

nevertheless the increase in resistance due to it can be very large, part cularly wlth the high-inductance coils considered here. According to my investigations the hysteresis loss per second in the case of weak sinusoidal magnetizations up to about 10 (J. G. S. lines of force per square cm. is proportional to the cube of the amplitude of the induction density Such 1 and also proportional to the frequency. The increase in resistance R2 in ohms due to this loss can be expressed by the following formula: R2=As Cf, where A is a constant to be determined experimentally, s is the number of turns in the coil-winding, O is the magnetizing-current in ampcres, and f is the frequency of the current. The experimental constant- A depends on the size and shape. of the iron core. I have found that in the case of coils considered here, in which the core is laminated suitably, so that R1 is about one-tenth of the true resistance of high-resistance telephone-cables and about equal to the true resistance of low-resistance air-lines fora frequency of 1,500 p. p. s., then for currents up to five milliamperes and a frequency of 1,500 p. p. .s. Rz Wlll be a small fraction of R1 as long as the number of turns per centimeter of the shortest magnetic line of force is not greater than 60. This much will suffice to give a practical approximation of the constant A and to enable one skilled in the art to adjust the size and number of the iron plates of the core and the, number of turns in the winding.

To take the above example, let there be 666 plates. Since each one is .0075 cm. thick, we shall have in the formula for L the product 42 (Z equals 5. Let the internal diameter of each plate be 1 cm. and the external diameter 8 cm.

Hence log 6 I (approx) Finally, since L equals .2 and. 1 equals 10 we shall have .2 10 2X 82X 10 5 X .7. Hence .9 equals 535 turns.

The shortest magnetic line of force. is a little more than 12 cm. Hence there will be less than 60 turns per cm., and the increase in resistance dueto hysteresis will be, up to. 5 milliamperes and 1,500 p. p. s. a small fraction of 8 ohms,

-In the construction of the cores for. coils (represented by Figs. 1, 2, 4:, and 5) iron wire can be employed in place of iron plates. Prac-- tically the same formulae and the same general :remarks will apply to iron-wire cores; but it that in the expression for this inductance L the factor and must be replaced by the actual height of the core.

Having now described my invention in the best way known to me for practicing the same, what I claim, and desire to secure by Letters Patent, is

1. An inductance-coil having an iron core,

I'IO

forming a closed magnetic circuit, in which the resistance due to hysteresis is smaller than the resistance due to Foucault currents for current strengths and for the highest important frequencies (about 1,500 p. p. s.) in telegraphy or telephony, substantially as described.

2. An inductance-coil having an iron core, forming a closed magnetic circuit, in which the resistance due to iron-core losses of energy is smaller than five times the true ohmic resistance of the winding of the coil for all telephonic frequencies up to 1,500 p. p. s. and for current strengths of importance in telegraphy or telephony, substantially as described.

3. An inductance-coil consisting of an iron core, forming a closed magnetic circuit and the winding of which is distributed over this core uniformly and symmetrically, the core and winding being adjusted in such a way as to give the coil for a predetermined frequency and current strength a predetermined inductance and resistance, substantially as described.

4. An inductance-coil consisting of an iron core, forming a closed magnetic circuit, the winding of which is distributed over this core uniformly and symmetrically, the core and winding being adjusted in such a way as to give to the coil for a predetermined frequency and current strength a predetermined ind uctance and resistance, and to give a resistance due to hysteresis, which is smaller than the resistance due to Foucault currents for the current strengths and for the highest important frequencies (about 1,500 p. p. s.) in telegraphy and telephony, substantially as described.

5. An inductance-coil consisting of an iron core, forming a closed magnetic circuit, and a winding which is distributed over this core uniformly and symmetrically, the core and winding being adjusted in such a way as to give to the coil for a predetermi ned frequency and current strength a predetermined induct ance and resistance, and to insure that the resistance due to iron-core losses of energy is smaller than five times the true ohmic resistance of the windings of the coil for all telephonic frequencies up to fifteen hundred pulsations per second and for current strengths of importance in telcgraphy and telephony, substantially as described.

(3. A twin inductance-coil consisting of an iron core, forming a closed magnetic circuit, over which the winding is distributed uniformly and symmetrically, the core being adjusted in such a way as to give to the coil for a predetermined frequency and current strength a predetermined inductance and resistance, substantially as described.

7 A twin inductance-coil consisting of an iron core, forming a closed magnetic circuit, over which each winding is distributed symmetrically and uniformly, the core and the windings being adjusted in such a way as to give to the coil for a predetermened frequency and current strength a predetermined inductance and resistance, substantially as described.

8. In a system for electrical-wave transmission, a non-uniform conductor, consisting of a uniform conductor with inductance-coils, each consisting of an iron core forming a closed magnetic circuit, over which the winding is distributed symmetrically and uniformly, connccted in the uniform conductor at periodically-recurring points, the inductance, resistance and capacity of the interposed ind uctancecoils being adjusted in such a way as to give for the highest frequency to be transmitted, with the inductance, resistance and capacity of the uniform conductor, a prraletermineil inductance, resistance and capacity per unit length, the resistance due to iron-core losses of energy in the coils being smaller than the true ohmic resistance of the loaded conductor for the highest frequency to be transmitted substantially as described.

9. In a system for electrical-wave transmission, a non-uniform wave-conductor, consisting of a uniform conductor with twin inductance-coils each consisting of an iron core, forming a closed magnetic circuit, over which the winding is distributed symmetrically and uniformly, connected in the uniform conductor at periodically-recurring points, the inductance, resistance and capacity of the inter posed inductance-coils being adjusted in such a way as to give for the highest frequency to be transmitted, with the inductance, resistance and capacity of the uniform conductor, a predetermined inductance, resistance and capacity per unit length, substantially as described.

10. In a system for electrical-wave transmission, a non-uniform conductor, consistingof a uniform conductor with twin imluctance-coils, each consisting of an iron core formi n g a closed magnetic circuit, over which the winding is distributed symmetrically and uniformly, connected in the uniform conductor at periodically-recurring points, the inductance, resistance and capacity of the interposed iiuluctanew coils being adjusted in such a way as to give for the highest frequency to be transmitted, with the inductance, resistance and capacity of the uniform conductor, a predetermined inductance, resistance and capacity per unit length, the resistance due to iron-core losses of energy in the coils being smaller than the true ohmic resistance of the loaded conductor for the highest frequency to be transmitted, substantially as describe].

Signed by me in New York city, (borough of Manhattan) New York, this 5th day of February, 1903.

MICHAEL I. .lUllN. Vitnesses:

JULIA MURPHY, GEORGE H. GILMAN.

Images