US1749045A - Signaling system - Google Patents

Description

Patented Mar. 4, 1930 UNITED, STATES' PATENT OFFICE HARRY NYQUIST, 0F MILLBIURN, AND KENNETH W. PFLEGER, 0F ARLINGTON, NEW JERSEY, ASSIGNORS TO AMERICAN TELEPHONE AND TELEGRA'PE COMPANY, A

CORPORATION OF NEW YORK Application filed Jinic 11,

It is an object of our invention to provide for effectively transmitting a composite current comprising component currents of frequencies in a band from zero up to a considerable value. Another object of our invention is eectively to transmit the lower end part of such a band of frequencies by cutting, it ofi and stepping it up in frequency above the upper end of the band for actual'v frequency somewhat above said limiting value. Still another object of our lnvention is to-separate the components of a signal into separate channels according to their frequency above and below acertain critical intermediate frequency, and then to recombine them with proper attenuation equali- `zation vand. phase correction so that as recombined `the resultant signal wave will correspond accurately with the initial signal wave. All these` objects and other objects of our invention will be made apparent in the following specification and claims taken with the accompanyin,gr drawings, in which we have disclosed an example of the practice of our invention. yIt will be understood that this disclosure is specific to this example, and that the scope of' the invention is intended to be indicated in the appended clams. V In the drawings, Figure l is a diagram of apparatus at the transmitting end of a signaling system;.l `i f. i2 is a diagram of appa- 4ratus at thereceiving end; and Figs. 3 to 6 are coordinate diagrams of amplitude and phase shift as functions of frequency.

' To illustrate our invention, it will be.

shown how an increase in speed of telegraph transmission may be obtained over a telephone line whose band of transmitted frequencies is limited. This inventionlis appli- SIGNLING SYSTEM i 1927. seran m. 198,285.

ing a total transmitted band of l800 cycles in width. If the speed is doubled, then a line would be required which transmits twice that band, or 1,600 cycles. If the speed is again doubled, a line would be required which transmits twice 1,600. or 3,200 cycles. This assumes, of course, that sending and receiving mechanisms can be devised to operate at these speeds. The latter speed cannot be handled by medium-heavy loaded circuits since they cut of at about 2,800 cycles. Neither is it possible to transmit the envelope of theV carrier wave as a series of D. C. pulses ranging in frequency components from zero to v 1,600 cycles because the line and apparatus do not transmit the very low frequencies. But it is possible to obtain this highest speed with such a line when the very low-frequency components are separated from the rest of the signal and transmitted by carrier waves. Figs. 1 and 2 show an application of this method to the transmission of a signal including frequencies from zero to 1.500 cycles. This range is indicated in Fig. 6. It is assumed that the cable does not transmit frequencies much below 200 cycles. ,The filters located at A and B in Fig. 1 divide the original signal into two paths,-one transmitting .from zero to about 200 cycles and the other from about 200 to 1.500 cycles. The currents of the/path A are too low in frequency to be transmitted over the cable and consequently are used to modulate an 1,800-,cycle carrier wave, the products of modulation being about 1,600 to 2.000 cycles. which are easily transmitted by the cable. The currents from the two paths a-re combinedand transmitted over the cable, the frequencies ranging from 200 to 2.000 cycles.

A carrier Wave of 1,800; cycles is used inlthan 200 cycles, while in path stead of 1,700 cycles (which would give 'prod- .ucts of modulation from 1,500 to 1,900 cycles) in order to allow for the fact that the filters cannot be made to cut off sharply and to avoid overlapping of the lo'wcr side band with the frequencies below 1,500 cycles. This 'is made apparent in Fig. 6. 'At the receiving end ofthe line the ilter's in paths A and B" in Fig. 2 separate the products of modulation from the other frequencies. The former pass through-a detector and reappear at the output as currents of zero to 200 cycles. Transmission equalizers are shown in order ,to adjust the amplitudes of -the currents ofeackpath to the proper values so that on combinlng them the original signal will be-exactly reproduced, provided the currents are all in the correct phase relation. The details of construction of transmission equalizers, phase equalizers, and filters are well lmown and therefore they will not be given here.

The correct phase relation for distortionless transmission is'obtained Joy means of.

phase equalizers which may be separate from the transmission equalizing networks. Such phase equalizers may consist of lattice type all-pass networks or their equivalent. Due to the fact that the filters used in Fig. 1 do not cut oil'I sharply at 200 cycles it will befound that in the path A of Fig. 2, currents will exist of frequencies sli htly greater there will be some frequencies less than 200. cycles. It is necessary that the phases of currents-in this overlapping portion of the frequency band beso adjusted that 'where paths A and B join, the currents will not cancel due to being in phase opposition. In choosing the transmission and phase equalizers of Fig. Y 2, it is necessary to know what-phase displacement and amplitude each current -in path A has at the output of the detector and likewise in path B at the output of the low-pass filter. This" information may be obtained either by computation or by measurement after the circuit has been set up. In order that the combined currents should give thev exact reproduction'of the original signal it is first necessary that currents of all frequencies, 04,500 cycles be transmitted e ually. ,'lherefore, the transmission equalizers Nos. 1 and 2 should be arran ed to -correct the amplitudes according to ig. 3 so that when currents of the two paths are combined in phase the resulting amplitude-frequency curve will bea straight line, equal to the sum of the two curves in F ig. 3.

Thephase of currents in paths A and B measured before correction may be plotted as two separate `curves shown in Fig. 4 which may o' most likely', may not be straight lines. Unies e. curves-coincide over the frequency `,tmmnuori@te'.jiboth paths or are parallel and u' lmultipleet 360o apart then the two multiple of 360 apart throughout this range.

The system disclosed. will be dependent for success on the design of these networks. Perhaps the simplest networks to construct are ladder-type filters. One or more sections of simple low pass filter, with series inductance and shunt'capacity, may be used for the phase equalizer of path'A in Fig. 2f The phase shift per section in the non-attenuation range, is given by the formula:

Sm f. where fc=cutoff frequency of the filter. The phase shift is' zero at zero frequency and increases faster ,and faster with increase' in frequency up to the cut-0H point. By locating, f., above 250 cycles the desired phase shift may be obtained (concave upwardlwhich,

added to the curve for path A of Fig. 4,-

gives the curve for path AUof Fig. 5. The number of sections and the location of fc for each' section is an adjustment to be determined bythe designer cfa particular installation.y or example, by raisin can be made less concave, whilegby increasing the number of sections the curve can be made steeper, and the total phase shift increased.

In calculating the total phaseI shift caused Joya` a number of sections of filter it is, of course,

ln'ecessary, for accurate results, to take account of the impedances connected to the filter.

The phase Aequalizer' of-pa'th B in- Fig. '2 may consist of oneor more sections of simple c, the curve 'high pass filter, with series capacity and shunt'inductance. The phase shift per section in the' non-attenuating range is given by the formula: f

The phase shift is Hzero at ,infinite fre uency and increases negatively with decrease 1n frequency down to the cut-oil' point. B locating fc below 150 cycles, the desire phase shiftvm'ay be obtained (concave downward) which, when added to the curve for path B of Fig. 4 gives the curve for path B of the location of the respective cut-off Afrequencies is determined by the designer for each installation just as in the case ofthe low pass filters.l It is also possible to accomplish the desired result by the use of latticetype all-pass networks or their equivalent Fig. 5. The exact number of sections and I which-have the advantage of possessin a constant characteristic impedance lat all requencies, which simplifies the calculations when they are terminated in matchedjmpedances. .In this case it is only necessary to consider the equation of each section which givesthe phase shift: l ,v

.'B with theirfo above 200 cycles. By taking the `right number of sections and adjusting the constant I) of` each network, the slopes of the resulting phase curves for each path may be varied until theycome tan ent or meetthe l specifications outlinedabove or proper joining ofthe paths. In'order to obtain a180 phase shift whenever necessary the connections to. one of thep'aths maybe reversed.

For accurate work it is necessary, of course,

that the phase and transmission equalizers be designed to allow for the fact that in practical'construction the former will have some effect on the attenuation and thelatter some v eft'ecton the phase.

paths A and B will then have the desired amplitude characteristic but probably a somewhat irregular delay curve. The phase equalize'r directly on the input side of the receiver in Fig. 2, consisting of lattice-type networks, may e employed 1n the customary manner, to give a flat delay curve-to currents entering thereceiving device. When'all frequencies are transmitted' with substantially equal lamplitudes and delay,'the signals vare practically ree from' distortion. e r o Weclaim: p e ,l l 1. The method ofy transmitting a signal,

A,which consists in separating its components into respective channels approximately above and below a certain critical frequency, equalizing the attenuation and equalizing Vthe phase shift in these channels to offset de- `.partures occasioned by said separation, and then recombining "the ycomponents to produce a resultant signal like 4'the initiall signal. y

2. The method ofcompensatingfor the ldeparture from, exact cut-off `given by filters when'separatingomponents of a' signal wave' 1n separate channels according to frequency,-

whichconsists in equalizing the attenuation andpha'se'shift in the separate channels and thereafter combining the currents in those channels. .A

` 3. In combination, a channel, .branches therefrom, means to effectan approximate separation of the components of a signal current wave from said channel into said branches according to frequency, means in e said branches to compensate for attenuation andphase shift, and kmeans, to combine the components in theA branches whereby the resul'tantfcurrent will have` the same signal y form as the initial current. 14. The method of separa-ting components of a signal currentaccording to frequency and recombining them, which consists in iil- ,tering such components apart withso much distortion at frequencies near theseparation 'frequency as may be unavoidable, and ythen ycompensating the separated current compovnents forattenuation and phase shift so that when recombined the resultant current will correspond to the initial current.

. 5.. In thel method of combining currents of adjacent frequency ranges with slight overlap, the step which consists in compensating themfor attenuation and phase shift at frequencies which overlap, so that when they are combinedthere lwill be no irregularity of amplitude nor of phase shift at those frequencies.

6. Incombination, a line'good for transmitting currentsv of frequencies only from a tod, means at the sending end to enerate signal currents with components at requencies from 0 tol bv, means to separate these into two channels with ranges from 0 to a and t y lfrom a to b, means to generate a carrier at The combined currents at the junction of c and modulate it by the said low'er range, thus producing a modulated output of range within ca to c-l-a, means to apply such output and the current of rangen to b to the'said line, means toseparate these ranges at the receiving end, a detector to getv the current components from O to a froml the higher'received range, compensators for attenuation and for phase shift for the sepa-` `channels below and above a, stepping the lower'such range up to a range lying between b and d, then transmitting both ranges on said line, and, ,at the: receiving end, separating them and stepping the higher range downto 0 to a, compensating these separated ranges for attenuation and phase shift and combining them'to produce a sicnal wave like the initial wave at the transmitting end,

the frequency. values having magnitudes in the order 0, a, b, anddwith d-"b 2a.'

8. -The method of effecting transmission of signal currents-of component frequencies from zero up to a considerablevalue over a line which Will notv transmit' directly currentsv of low frequency, which consists in y separating off thelow frequency components,

stepping them up in frequency above the upper limit of the signal'frequency frange, transmitting both ranges, and, at the receiving end,'stepping the higherrrange down to the corresponding initial values, compensat-` ing in bothran'ges for attenuation and phase and thenrecombining the components to produce a resultant'signal like the initial signal.`

In testimony whereof, we have signed our names to this specification this 9th day of June, 1927. y

HARRY NYQUIST. KENNETH W. PFLEGER.

shift and combining and applying therethe said last-mentioned range and the said second range, and at the receiving end, separating the second range and third range currents, stepping the third range currents down to their Yinitial values as in the-first range, andthen combining the resultant currents to get a signal currentcorresponding to the initial signal current at the sending end.

l0. In combination, means to produce a signal current having frequency components lying in two ranges which may be designated first and second, a line adapted to transmit frequency components in the second rangeand another range which may be designated .as the third, means at the sending end to separate the components of the first and-second ranges and to shift the frequency of the first range components into vthe third range, means to apply the currents of the second and third ranges to the line, and. at the receiving end, means to separate these ranges, means to step the currents in the third range back to the initial' values of the first range, means to compensate the first range currents and the second range `currents for attenuation and phase shift, and means to combine and applythe currents so compensated to produce a resultant signal wave like the initial signal Wave at the sending end.

llt; The method of transmitting a signal, which consists in separating its components into respective channels approximately above and below a certain critical frequency, equalizing the attenuation and equalizing the phase shift in each of vthese channels to Oifsct departures occasioned by said separation,

Images