US1679434A - Electric wave filter for variable-load circuits - Google Patents

Description

Aug. 7, 1928.

R. G. M CURDY ELECTRIC WAVE FILTER FOR VARIABLE LOAD CIRCUITS Filed April 21. 1926 2 Sheets-Sheet file 0M4: [we v L Liza Paws/lake for railway squads file whom? Line fill/er line INVENTOR ATTORNEY Aug. 7, 1928.

R. G. M CURDY ELECTRIC WAVE FILTER FOR VARIAB LE LOAD CIRCUITS Filed April 21, 1926 2 Sheets-Sheet INVENTOR 1?. 6f JfiCara y ATTQRNEY Patented Aug. 7,'1928.

UNITED STATES PATENT OFFICE.

RALPH G. MOC'URDY, OF ENGLEWOOD, NEW JERSEY, ASSIGNOR TO AMERICAN TELE- PHONE AND TELEGRAPH COMPANY, A CORPORATION OF NEW YORK.

ELECTRIC WAVE FILTER FOR VARIABLE-LOAD CIRCUITS.

Application filed April 21, 1926. Serial No. 103,615.

An object of my invention is to provide "a new and improved system for suppressing Another ob-' harmonics on a power circuit. ject of my invention is to provide an electric wave filter to be interposed in a line without affecting the load impedance on the source for the current of frequency desired to be transmitted. Another object-of my invention is to suppress harmonics on a power line that parallels a telephone line without altering the impedance which the power line presents to the source supplying'it. These objects and various other objects of my invention will become apparent on consideration-of a limited number of examples of practice according to the invention which I have chosen here to disclose to illustrate the nature and use of the invention. It will beunderstood that the present specification relates largely tothese examples of the invention, and that the invention will be defined in the appended claims.

Referring to the drawings, Figure 1 is a diagram of a power line paralleling .a telehone line, and equipped according to my. Invention; Fig, 2 isa detail diagram of the filter shown symbolically in Fig. 1; Fig. 3 is a diagram of a 3-phase system corresponding to the single-phase system shown in Fig. 1; Fig. 4 is a detail diagram of the filter for the 3-phase system shown symbolically in Fig. 3; and Fig.5 is a curve diagram to show performance in a certain case. I

As an example of practice under my invention I show in Fig. l a telephone line 21 extending several hundred miles, and in parallel therewith a power line 22 for the operation of railway signals along the rightot-way. The current for thlsline 22 is of cycles from the generator 23, and in the particular example that I have inview, it is put on the line 22 at as high as 4,400 volts. The load on the line varies over a rather wide range from time to time, and this is indicated symbolically by the loads 24 with switches 25, by which they may be temporarily connected across the line 22. Under some conditions the power put on the 'line 22 from the generator 23 may be as high as 21.5 kilovolt-ampercs. I

Obviously. there will be a considerable tendency to induce currents in the telephone line 21. So far as currents of the fundamental frequency of 60. cycles per second are concerned, these are below the essential methods they can be eliminated or ya ueC/2 in voice frequency range, and by well-known compensated or otherwise dealt with effectively in the telephone line 21. But it will ordinarily be the case that harmonics will be present, especially odd harmonics, in the output'from the generator 23; These harmonics, being of higher frequency, Will be comprised in the essential voice frequency range and will introduce complexities if they appear in the telephone line 21, and be difficult to eliminate or compensate or deal with effectively in that line.

Filters are ordinarily designed for stable impedance conditions in the load connected on their output sides. In an ordinary filter, if the impedance connected to the output should vary, the electromotive force across the output terminals would also vary even though it were kept constant on the input side of the-filter. Hence the introduction of an ordinary filter, as at 26 in Fig. 1, is impracticable, for, as the load varies on the line 22, it would cause a variation of electromotive force on the line even though the generator 23 is of the usual constant voltage type. I have divised a filter that will suppress the harmonics but will transmit the undament-al, and such that the impedance looking into the filter across its input terminals is always substantially the same for the fundamental frequency as the impedance looking into the line from the filter terminals, no matter how the load on the line may vary.

A articular filter that is adapted to be intro need, at 25 is shown in Fig. 2. This is a low-pass filter, consisting of two midshunt sections. That is, there are two sections, each of a type, and each consisting of a series inductance L and two shunt capacities in the two 'le s each of value C/2. In F i 2 the two afjacent capacities, each of parallel, are united in the single capacity C.

It is a feature of design of the filter shown in Fig. 2 that its cut-ofi frequency is /27 where f is the frequency that-is to be passed; 60 cycles in this case. This cut-off frequency is related to the values L and C by the wellknown equation for the low-pass filter,"

mesa

The foregoing equation imposes one condioutput tion of design on the coil L and the condcnser C. Another condition lies in making them of such values that they will safely transmit currents as high as to give kilowatts with unity power factor at 4,400 volts. along in this specification, this leads to the assignment of impedance values at 60 cycles as follows: for each of the coils L, +j 900 ohms, for the condenser C, 450 ohms; and for each of the condensers C/2, j 900 ohms.

With these impedance values for the elements of the filter, and with an impedance of Z connected to the filter input, and Z connected to the filter output, and with an electromotive force E applied at the filter input, the complete data are furnished for solving the network of Fig. 2, and it will be found that the current 1 through the terminal impedance Z,, is given-by the equation,

The result given in Equation (2) may be obtained by formula16 on page 284 of K. S. Johnsons Transmission Circuit-s for Telephonic Communication, published by D. Van Nostrand, New York. Thus it will be seen that for current of 60 cycles the interposition of the filter in Fig. 2 is of no effect on the magnitude of the current through the load. No matter What the value of the load Z maybe, the current will be the same as if it were connected directly to the source E through the source impedance Z.

Thus far the discussion has been based on the assumption that the coils and condensers of the filter are pure reactances. \Vith rea sonable care coils may be constructed to have a resistance value R, such that 21rfL 57 (3) The diagram of Fig. 5 is obtained subject to Equation (3). This is a logarithmic diagram, and with the legends thereon expressed it shows that as the load varies from 0.1 of full load up to 1.1 times full load, the value of the voltage E across the filter output terminals remains very nearly constant. All the curves of Fig. 5 are drawn for unity power factor. The diagram also As Will be shown a little farthcn cycles. the attenuation is as high as 37 or 38 T Us.

The particular installation that I have in view as an example of practice under my invention is on a single-phase power line, for which it may be desirable. later to add another conductor and make it S-phase. Therefore, the installation is made as shown in Fig. at, where each inductance L is represented by two coils each of value L LZ, and the condenser C is represented by two condensers in series, each of value 2C. and each condenser C/Q of Fig. 2 is represented by two condensers in series in Fig. 4 each of value C. Z

If. at a later stage, the system is changed to 3-phase. it will then be necessary only to add the coils and condensers marked with primed letters. and to make the changes of connection that will be obtained by throwing the three switches S and the two switches S.

For any periodic artificial line of 11 sections, starting at mid series or mid shunt, the following formulas hold:

sinh n I=0 (7) assuming Z, and Z are finite and not zero, which is obviously a proper assumption. When Equation (7 obtains, it follows that cosh n I=i1 and tanh n- F=O, and Equations (4), (5) and (6) reduce to E,/E 1 /1 i1 and Z,/Z,=1.

The solutions for Equation (7) are given y nF=j7c1r (k=0, 1,2. (8)

Let the periodic artificial line he of ladder-type with series impedances 2 and shunt impedances 2 For such a line -1 I- cosh (1 22 (9) To arrive at the filter shown in Fig. 3

cosh r= h eef Lo '(10) is the frequency. I Let the critical or wheref be f for cut-0d frequency of the filter Which Now consider the frequency f such that J" fo/. Substituting this value in Equation (10), and by the aid of Equation (11), welget that cosh -I'O, and I=j From this we see that an even number 210 of sections of the network considered will give the equation I 2kI=jk1r.

This satisfies'Equation (8) above when n. of that equation is'an even number. More-. at (Equation (9) on over, the characteristic impedance f' fo 2 is found to be page 125 of the said Johnson book) I which'is finite and hence the conditions for 1' u ttenuated transmission at all-loads are fu filled;

/ In the design of the filter, the characteristic impedance is determined from the equa- I where P is the power. In the present case, letting P.=21.5 kilowatts and E=4,400 volts, we get the value of Z at 900 ohms. Substituting this in Equation 12), and assigning a chosen value for f in Equation (11.), namely f.,=6O /2, it is seen that these two equations (11) and (12) serve to fixthe values ofL and C.

The condition of perfect regulation is obtained when the phase shift through the filter for the frequencies to be transmitted is 180" This maybe obtained by using a number N of sections, the requirement being that the phase shift through one section be 0 e i? The number of sections inust be in,- creased when the cut-off frequency is increased, approaching infinityas the line approaches a smooth line. By properly choosing the number of sections, anydesired cutoff frequency from-the square root of two times the transmitted frequency maybe approximated. In the foregoing specific example of my invention, it will be 'seenthat I have chosen to make the filter of two secgin'.

tions and the cut-off frequency times the fundamental frequency.'

After the foregoing Equation (9) we assumed a low-pass filter; but We might, for example, assume a confluent band pass filter with mid-section terminations and we should find that if its critical or cut-ofi' frequencies were chosen so that the frequency to be passed was a mean proportiona between them, then it would transmit freely for all loads at the frequency to be passed.- In this case, a reasonable assumption would be that the upper cut-off frequency is double .the frequency to be passed; this would out off the lowest odd harmonic by a safe mar- This assumption easily ,leads to the condition that each series ,coil shall have 8 times the inductance of each shunt coil. I' claim:

1. Awave filter-to pass current of a certain frequency of, with substantially zero impedance change for current of said frequency through the filter notwithstandin changes of the output impedance, said fi ter being of lowpass ladder type, and consisting of two dshunt sections and having'its cut'oflI' frequency at times said certain frequency.

2.. A network to be interposed between an alternating current source of constant voltage and a variable impedance load to pass current of fundamental frequency f and to stop harmonics thereof and to keep the network input impedance substantially the same as the load impedance, said network being of laddertype with three cross memand to stop harmonics therei bers of successive capacity values 6/2, C and (3/2, and two alternately disposed series members each of inductance value L, subject to the condition thatf=1 1r VzLO,

3. A wave filter to pass current of a certain frequency and to stop harmonics thereof, with substantially zero impedance change for current of said frequency through the filter notwithstanding changes of the output impedance, said filter being of low-pass ladder type, and consisting of an even number of sections and having its cut off frequency at 2times said certain frequency;

4. A wave filter to pass current of a certain freqn'enc and to stop harmonics thereof, with 1 s bstantially zero impedance change for current of said frequency through the filter notwithstanding changes of the output impedance, said filter being of ladder type withinid-section terminations and With a pass range of frequencies comprisingsaid certain frequency and with the A critical fre- .quencies at the end,of the pass range so reom a constant voltage source the method of suppressing harmonics without i ntroducing substantial sion of frequencies u certain frequency an above that value, an

to 2 times the said cut-off of frequencies d with phase-relations throughout the filtering process to keep the input and output at the desired frequency in phase or in phase opposition.

In testimony whereof, I have signed my name to this specification this 20th day of April, 1926.

RALPH G. MGCURDY.

Images