US1666206A - Modulation system - Google Patents

Description

April 17, 1928.

1,666,206 R. v. L. HARTLEY MODULATION SYSTEM Filed Jan. l5, 1925 /nvenars Rolo/7 L/L Hari/ey Patented Apr. 17, 1928.

UNITED STATES PATENT OFFICE.

RALPH v. n HABTLEY, or SOUTH ORANGE, `miur JERSEY, ASSIGNOR, BY RESINE AB- sIoNMENTs, To WESTERN nLEcTnIo oOMrANY. INCORPORATED, A CORPORATION OF NEW YORK.

MODULATION SYSTEM Application led January in usefully employed energy and asimilar economy of frequency rango.

Herctoforc, it. has been necessary first to generate a. double side band modulated car rier wave and then to select from` this wave, as by means of filters, a portion containing a single side hand. Since the present invention relates only to the production of a single side band Wave, without regard to the presence or absence of an unmodulated Carrier component, a similar object will logically bc assumed in considering the prior method. The problem, then, has heen se arate by this filter means, closely adjacent reqiiencies. The difficulty in accomplishing this is inversely proportional to the ratio of the frequency separation of the side bands to their meank frequency that is, to the carrier frequency. The diiiiculty accordingly in-` creases with increasing' carrier frequency.

Let it be assumed, as in the usual case, that the side bauds in quest-ion are produced by simple modulation, that is, hy a single modulatingrstep, and that the modulatingr frequencies are among those commonly used in signal transmission. For these conditions and using' the prior method of single side band production a practical limit of carrier frequency is reached, beyond which separation of the side band cannot be accom lished, which is less than the frequency or inarily required for radio transmission. This limit has been found to be of the order of 30,000

cycles. Accordingly for the production ofa single side band of a radio carrier Wave simple modulation and selection is not Sullicient.

In order to overcome this difficulty resort has been had prior to this invention to a.

method of multiple modulation. This method insures that for successive steps of carrier frequency increase, there will be a elaborate` and expensive outlay 15, 192i. Serial No. 2,483.

proportional increase in the spacing between the carrier and its modulated side bands so that there will be little ditiicult-y in pcrforming the necessary frequency separation after the last stage of modu ation. However, such a method is involved and requires an for circuits and apparat-us.

Itis an object of this invention to provide improved methods of and means for producing a modulated carrier Wave having a single side lband. Y

Itis a further object of the invention to achieve the above object` by methods and means which donot depend for their'oiiicacy on acritical value of either carrier or side band frequency or on a particular relation` of these frequencies.

The method involves the production of two pairs of side bands, using individual modulating,r circuits but the same carrier and modulating frequencies, and which pairs of side bands are so related that the phases of the current components in two side bands, one from each pair, are equal, and that the phases of the components in the remaining side hands arein opposition. y The two pairs of side bands thus related are superposed with a resultant balancing out of one side hand and a doubling of the amplitudes of the components in the other side hand.

The necessary relation of phases for the current com onents in the two pairs of side bands is achieved by the use of certain critical phasev relations between the carrier and modulating potentials that are impressed on the respective modulators. The relation may be satisfied, for example, by causing the phases of the carrier potentials for the two modulators to differ from nach other by ninety electrical degrees and similarly causing the phases of the corresponding modulating potentials to diiier hy ninety degrees. Thisis on the assumption that second order modulation, that is, modulation resulting in side bands in which the frequencies are indicated by mp plus and minus g, and in which m equals one, isused in each modu- Vlating circuit, p and q indicating respectively ther orders or methods of modulation resulting in side bands of the same typ; but in which m is greater than one may used, which may be the same or different in the two modulating circuits and which may require different relations of the phases of they im iresscd` potentials. (.)f coursez where other tran second'order modulation 1s used,

in one or more of the modulating circuits,

. the frequencies of the impressed carrier po- 'for achieving this result.

Vtentials must be caused to have the propel' harmonic' relation to cach other or to `the desired fre uency mp to insure that the frequencies wi l be the same for the two pairs of resultant side bands. Reference is made to U. S. application of Peterson, Serial No, 683,301, tiled December 29, 1923 for a complete analysis ofsecond and higher order modulation.'l 'lhc gene-ral solution of the method of the invention for producing sin le sido bands as well as several particular so ut of single sideband modulation system'has,`

however, waited upon the conception of an. operative means for producing this uniform phase shift for a band'of frequencies. Accordingly it is a still further object of the inventlen to provide methods of and means The method of providing 'a uniform relative phase shift may be practised without regard to whether the bands of frequency between which the` uniform phase snift is provided did or did not originally comprise portions of a single current. as-in the specific application of the principle herein disclosed. and without regard to whether or whethernot the correspending frequency cemponcl'its-in the two bands had originally the same fphase.`

The solution of the uniform phase shift problem proposed in this invention is of a ind which cannot well be treated otherwise than bv a detailed mathematical Vdemonstration. t must therefore suiiice at present to state that it depends on the fact that for certain filter networks the phase shift per section varies nearly uniformly throughout the transmission range and that the .slope of the phase shift-frequency curve is a function of the transmission range of the filter and of the number of sections. The modulating circuit is branched. Filters are-ineluded in the branch circuits. These filters differ" in on band widths and in sacrifice.

numbers of sections with the result that although in neither branch circuit does a uniform phase shift occur, a uniform relative phase shift occurs between the Vcomponents transmitted by the filters in the two circuits. A

The invention may now be better understood from the following detailed description thereof when read in connection with the accompanying drawin i in which Fig. 1 illustrates a preferred e bodiment of the invention and in which Fig.,2 comprises certain characteristic curves drawn artially to scale and which areused in exp aining the operation of the invention.

In the description of this system of the invention and its method of o eration with reference to the drawing certain results will be assumed without proof, or at least without uantitative dat-a. This omission will then e suppliedby mathematical demonstration wherein Vthe necessary conditions for eti'ectin a uniform relative phaseshift of a band o frequencies and their application 1n system for producing a single modulated side band will be developed.

Referring now to the drawing, the elements intermediate the circuits 1 and 2 cooperate to produce in circuit 2 a single side band of a carrier -Wave resulting from modulation of the wve from carrier source 3 by modulating waves from circuit 1. The conditions justifying a choice of frequencies for circuit 1 will be explained in more detail later.

The band of frequencies from. circuitl Vafter stepping up to a. desired value by combination with a single frequency wave from source 11 in modulator Ma is thenV transmitted through band pass filter ,Fr The purpose of thls filter, the reasonfor which will be made clear later, is to` confine the frequencies to be later used in modulating in modulators Ml and M2 to as narrow a band as possible consistent with thev grade of transmission desired. Thestepping up operation is reparatory to the subsequent operations. lthough, as will be explained later, this operation is conducive to greater accuracy in obtaining the desired uniform relative phase shi ft,it may be 'dispensed with under proper conditions Without serious These conditions will be pointed out later. The currents passed by filter F, tlow in branch circuits 5 and 6 which contain respectivelyband pass filters Fl and F,. In the particular branch arrangement illustrated, the terminations of the two filters are in series. An equivalent arrangement would result from putting these terminations in arallel, as is well understood. The currents ln these branch circuits are impressed respectively on modulators M, and M, in whlch they are combined-with carrier cnrrent from source 3. These carrier currents are amplified in amplifiers Al and .A2 respectively and impressed on the modulators through transformers 7 and 8.

In the particular circuit. illustrated it is assumed that second order modulation occurs in modulators MJl and M2. For this particular case it is necessary, in order to obtain the requisite relation of phases in the resultant pairs of side bands, that the carrier waves impressed ou the two modulators (litter relatively from each other by ninety electrical degrees. This result is achieved by impressing currents from the source 3 on amplifiers A, `and .A2 respectively across condenser 9 and resistance 10. The use of these two different types of potential impressing circuits insures the ldesired quadrature relation of impressed potentials.

Tt is also required, both for the assumed case and for other possible cases, that there be a ninety degree phase difference for each of the modulating frequencies in circuits 5 and 6. This is achieved by a proper choice of the number qt' filter sections and transmission ranges of filters I|`1 and F2. The particular theoretical relations governing the determination of thesequantities will be taken up under a separate heading later.

The various elements of the circuit, including the amplifiers and modulators, are cach old and of a conventional type. Accordingly. no further description of them is considered necessary.

Given the above relation of phases of thc frequency components impressed on the two modulators, there will result in the output circuits of these modulators two pairs of side bands. Two side bands, one from each pair, will have the same phase. The phases of the components in the other two side bands will be opposed. Accordingly by superposing these two pairs of side bands in circuit 2. one side band of each pair may be balanced out and the other side bands arithmetically added. This results in. the end desire. that is. the production of a single side band carrier modulated wave. Depending on the poling of transformers T1 and T2. the resultant side band may be the upper or the lower side band of the carrier wave. Be` sides the satisfaction of the precise phase relationship requiredV as above. it is also necessary that the amplitudes of the components of the two pairs of side bands be equal. This would ordinarilv result from using identical devices in the two symmetrical paths ot the system. However, the elements may differ somewhat from each other and the amplitudes may, in that case, be made the same by adjustments of the amplifiers and modulators as in accordance with conventional practice.

Although it was assumed above that second order modulation is used in the two branches, the principles of the invention do not exclude the use of higher order modulation in the two branches or second order in one branch and a higher order in the other. However. in order that the resultant side bands may have the desired frequency, care must he taken that the carrier frequency of the wave from source 3 has a proper suhharmonie relation to the desired frequency mp in the output circuit of a given modulator, if other than ordinary second order modulation is used in the modulator in question. Also. if side hands are produced of the type mp plus and minus (l in which m is different in thc cases of the two modulators, care must be taken that waves having the proper harmonic relation of frequencies are impressed on the two modulators. For example` it' second order modulation is used in modulator M2. and third order modulation is used in modulator M]` it would be necessary to impress on modulator M., the first even harmonic of the frequency of the wave. from source 3 and to impress on modulator Ml the fundamental frequency of the wave from that source. ln a `system so operated the amplifier A2 could be overloaded to function as a harmonic generator. For this specific system of modulation as well as for other modifications using other than second order modulation in each branch,V a phase relation of the impressed frequencies different from that described must be used. The conditions that must be satisfied in each of these cases. or in the general case, will be explained in detail hereinafter.

Theoretical basis for the met/rofl of producing a uniform relatrice phase shift for a band of frequencies.

This analysis will be largely based on the expositions contained in the following publications: a paper by Campbell` entitled Physical principles of the electric wave filter in the November. 1922. number of the Bell System Technical Journal; a paper by Zobel. entitled Theory and design of uniformand composite electric wave-filters in the January, 1923, number of the above journal; and U. S. patent to Campbell No. 1.227.113, granted May 22, 1917. yCertain ot' the formulae which are stated herein without proof are proved in some one of the above publications; the others are well known to those versed in the art and their proofs are omitted from this specification as being readily available from standard sources.

In any iterative network` of which a Wave filter is typical. the output current from a section is related to the input current to the section by a proportionality factor eAfB in which A-Iey'B is the propagation constant, A being the attenuation constant and B the phase angle (or phase constant).

For any ladder type filter, including the type disclosed in the drawing, the propagation constant B cos"(1 (1) It is proposed, starting with Equation (l), to first determine the phase angle-frequency characteristic of the type of filter disclosed in terms of the transmission range limiting frequencies and the frequencies f, at which the phase angle per section is E For the type of filter disclosed it may be shown that in which f is the given frequency and fl and Note that a is a function of the frequency interval between the frequency f., at which g and the variable frequency and that Z) is a function of the transmission range limits.

Substituting in Equation (3) II o 23 a3 25 3 a Suppose that there are nl and n: sections in the two filters F and F 2 respectively, nl being the larger. The subscripts introduced after this time Will refer to these respective filters.

the phase angle is The end to be achieved is to make B1-B2 inthis equation uniform for all frequencies within the band impressed from filter F3, to

f2 are respectively the lower and upper cutoff frequencies.

Therefore fzz-flzfaz-f3 The frequency at which the phase angle eqyualsll may be found from Equation (l) as follows cos 's l -l therefore, for the assumed phase angle in which y0 is the value of y for frequency fo.

'Yog l 2 This -value of yf substituted in Equation (2) gives Combining this equation with Equations (l) and (2) lgfwfz.

1t follows from the standard expansion B cos" `of the arc-cosine of an angle that Ef22 j12 Then II f (L 23 a. Bl-nlnyb grab-13- and l1 o 23 a3 0 B2 m2 -nzbz Gueb; y

The quantity a has the same value for both filters if fo is the same for both filters as will be assumed and as provided for in the practical `methods'of design to be cxplained later. The bs will have different values for filters of different transmission ranges, as will also he assumed for this case and as indicated by the use of different suhscripts. y

The relative phase shift will be elle 2 ripide] aibl 7l/2li 6a bia 623 a close first approximation. This end would be achieved if the terms in the equation containing a (that is, those which are a function of f) were equal to zero. There would then result a uniform relative phase shifty (n1-n2) Since n, and 'n2 are limited to integral values the shift would be limited to multiples of g that is, multiples of ninety electrical degrees. As a practical matter, since four ninety degree shifts, or a multiple thereof, would be the equivalent of a zero shift and since the other even multiple shifts could equally well be performed by transformers, an odd number of ninety degrec shifts would be sought, which would require the two filters to ditfer by an odd number of' sections. For use in the single side band production system described which requires a ninety degree shift, they would be caused to differ by one section. This condition will be assumed in the remaining part of the analysis.

It will be shown that the second term in the equation can be made equal to zero. Since the remaining terms, those involving the higher powers of a, have or can be made to have b proper choice of' filters, exceedingly small values, the result is a very close plroximation to the desired uniform relative siift.

The conditions accordingly are The solution of these equations would determine the principal features of the filters ff-fm and fs-f...

2 ELE fm faz 2 Equations (4) and (5) can now be solved for f,2 and f22 to give to give the desired phase difference in so far 40 as this difference is determined by the selfpro erties of the filters. However in a practic case the solution would require al consideration of the terminating impedances. For example if the terminating impedances, which are preferably resistances, are equal and if the filter im edances are nearly equal at the important frequencies the reflection effects will be substantially equal and will compensate each other. A trial solution will be made with certain constants more or less arbitrarily chosen and the phase shift and impedance curves will then be plotted for that case to check the accuracy of the solution.

In general it will be found best to choose f., well within the band of impressed frequencies and to choose the transmission range of the narrower` range filter at from three to four times the width of that band. The filter having the narrower range will be found to be the one which has the smaller number of sections, that is, filter F2. The transmission range limits must conform, for each filter, to the relation derived above or, for the respective filters,

fo2=fl22gf222 and fnggfna'iz'fzi" (4) y If fm equals the geometrical mean of the cut-off frequencies, so that fm2=f1 f2 the above formulae may be more conveniently used for determining the cut-off frequencies by substituting this value therein to give (fll al 2l)2 Then Let is be assumedthat the narrower range filter F2 has two sections and that filter Fl has three sections. v

Since, as explained above,

Since fo is the same for both filters 2 2 fi 15j-"a (1400er. (7)

Solving equations ((3) and (7) we obtain f21=l9350 c. p. s.

1:4180 c. p. s.

The important features of the design of' the two fillers is determined by the values of the constants obtained above.

In order to cheek the appropriateness of the choice of quantities more or less arbitrarily chosen the curves of Fig. 2 have been plotted. 1n this figure the abseiss are l'requcucies and the ordinates phase angles or impedances. 'lhe l'ull line curves are the phase angle eliannzterislics of the two filters F, and F2 as indicated by the labels l5, and B2 respectively. 'lhe dashed lines are the corresponding impedance characteristics, the curves K, and ,K2 corresponding respectively to filters F, and F2. 'lhe frequency rance of the impressed` frequency band is in icated along the axis of ahsciss. Since the filters are of similar type it follows `from the Campbell equations that equal phase angles per section correspond to equal impedani-es. Accordingly ,for the frequency at which the impedance curves cross, the phase angles per section are equal. But that frequency was chosen at which the individual phase angles Were frequency at which the phase difference is It therefore corresponds to the parrallel and separated by The nonlinearity does not appreciably affect the perellelism of the characteristics within the impressed frequency range for the reasons that this range is small relative to the transmission ranges of the filters and because it occupies a position near the center of those ranges. It is evident by a similar mode of reasoning that less favorable results would attend the use of a voice band itself, instead of a prepared, i. e., a stepped-up, voice band.

The impedance characteristics were determined from formul given by Campbell.

It is evident that the choice of range to be occupied by the impressed band and the choice of filter taken for the first trial were fortunate from a standpoint of uniform relative phase shift as affected by reflection since the impedanees are equal at frequency fo and nearly equal in that neighborhood and since this frequency f., was chosen within the impressed frequency band.

lt can he demonstrated with the aid of the references noted at the beginning of this analysis that, with a given band of impressed frequencies, as the number of sections, and hence the slopes of the phase an le characteristics increase, the reflection e ects due to the inequality of the impedances become less important. lf a sufficient number of seetions were used results comparable with those obtained in the example given could be obtained when using a voice band. 1n an actual case this would probably require fillers having at least twenty sections and differing, as in the ease given, by one section. 1t is apparent that the filter F3 may be made to contribute to this result by eliminating as large a portion of' the voice band at its lower edge as may be practically possible so as to permit as great a ratio of transmission range of the narrower filter to the range of the impressed frequency band as possible. It may contribute in a similar way, where a stepped-up band is used by limiting the width of this band to as small a limit as is practically possible for intelligible transmission.

'lo the extent that the signal band, for example, a voice bami, could be used directly, the method would replace the more complicated and expensive method involving nmltiple Asuccessive modulation. Where the speech band width is too great to permit the direct use of the method, as perhaps in the practical case, the method as applied in the manner illustrated in Fig. l, might still be useful as a step in side band elimination for very high carrier frequencies. For instance, suppose that it were attempted to build up a very high frequency single side band carrier modulated wave. Suppose that the final carrier frequency were so high that even after a preliminary step of modulation the finally produced side bands could not be separated by Selective means. The method of this invention could be practiced on the modulated wave resulting from this preliminary step so as to produce the desired single side band, since the width of a side llll band resulting from the preliminary step of separation of the side bands by selective means. By the use of the prior method, au additional step of modulation, making three in all, would have to be used. In the operation of this modification of the invention,

the system of the figure would accordingly replace the circuits required for the last two steps of modulation in a triple modulation method. For example, the method described in the U. S. patent to Espenschied 1,361,522,

granted December 7, 1920 could be modified in this manner.

The physical principle of the method can now be stated very simply in terms of the necessary steps required to make the hase angle characteristics of two filters di erent and parallel. If the two filters had the same transmission range and had the same number of sections obviously the two characteristics would be coincident. As the first step in making them different and parallel an odd number of sections, for example one, may be added to one of the filters. This would separate the curves but would increase the slope of the curve corresponding to the longer filter so that the curves would not be parallel. To compensate for this increase of slope its transmission range would be increased to produce a decrease in slope. Obviously for any excess number of sections in one filter there would be a critical value of transmission range of the other filter, narrower than thatof the first, which would tend to make the curves parallel.

Tiemet-cal basis for he 'method of' producng a single sn'e btt/ml 5 1/ bala/m1?.

The presentation of this theory, will, in general, follow that used in United States patent to Carson 1,449,382, granted March 2T, 1923. The treatment presented in United States application of Peterson 683,301 filed December 29, 1923 may be referred to with benefit in connection with the discussion of methods of modulation of other than the conventional second order.

According to the method used in the above mentioned patent, a statement of the potential (or current) resulting from modulation is obtained b substituting in the general equation of tie type y.:amfbH-cm values of the simultaneously impressed voltages (or currents). Suppose that the input potentials impressed on a modulator of the present system are P cos p, t and Q cos g, t in which p, and g, equal respectively 2a-p and 2a in iwhich p and g represent respectively t e carrier and signal frequencies, and P and Q, represent the corresponding maXi-/ mum values. Accordingly in the general equation equals P cos p, t-i-Q cos g, t. (No material chance would result if an initial phase angle between the two impressed waves were assumed.)

This vaille of should be substituted in the general equation. The first term, am, yields merely amplified waves of the impressed frequencies p and g. The term Zim2 yields waves of frequencies 2p, 2g, p-rg and p-g as is well known, the instantaneous values of the side bands p-i-g and p-g being represented by Z P Q cos (p,|g,)t and b Q cos (pf-gt. The frequency significant portions of t ese equations result from the trigonometric ex ansion of the product cos p, t cos g, t. I the other even power terms of the general equation are algebraically expanded by the binomial theorem and trigonometrically transformed as pointed out in pages (10) and (12) of the Peterson application Vabove mentioned, it will be found that additional side bands having the same frequencies are obtained which are superposed upon those derived from the sec` ond power term of the general equation.

Similarly to the above, from the odd power terms of the general equation, third order side bands of frequencies 2p+q and 2p-g may be obtained. These frequencies result from the trigonometric expansions bf the product eos 2 p, t cos y, t. This is cxplained in pages (l2) and (13) of the above mentioned application.

It could be shown in a manner similar to that used in the derivation of the above relations that side bands of other orders can be obtained from the general equation, even order side bands from the even power terms and odd order side bands from the odd power terms.

Assume now that the phases of the poten` tials impressed on the other modulator differ uniformlyfrom those correspondingly impressed on the first modulator. Let the phase difference of the carrier frequencies equal li and the phase difference of the signaling;l frequencies equal C. Let the quantities cos (p, +B)-and cos (q, t+C) be used iu place of cos p, t and cos 1, t in the mathematical operations above.

Considering first the steps that would yield second order side bands, the frequency significant portions of the quantities indicating the instantaneous values of these components, resulting from the trigonometric expansion of the product ly'have the value Vhether theluppcr or the lower side bands are balanced out is determined by the polingof the transformers T, and T2. It is obvious that the result would be the same if one or both the phase shifts B and C were in the other direction from the one assumed, or if one or the other shift occurred with respect to the other modulator. In fact, the only condition that needs to be specified, so far as the phases of impressed potentials are concerned, is that there must be a uniform relative phase shift of the modulating potentials and a relative phase shift of the carrier potentials.

For the case of third order modulation in each modulator, similar relations would govern between 2B and C. 'Ihat is, B must equal I and C must equal For this modification, the original carrier having frequency p must of course be the first even subharmonic of the desired carrier which corresponds to the generated side bands, and B is the phase shift between the original carriers.

If second order modulation is used in one modulatoi` and third order in lthe other, a carrier frequency 2 would be used in the second order inodu ator and the first even sub-harmonic thereof, p. in the third order modulator. The required relation of carrier frequencies for the two modulators could be obtained by using a harmonic generator in the circuit between the original carrier source of the second order modulator. Harmonic eneration may be accomplished very simply iy overloading the am liiier A, or A2, as the case may be, in accor( ance with the method described in United States patent to Kendall 1,446,752 granted February 27, 1923. The relation of phase shifts would be identical for this case as for the case where third order modulation is used in both instances. In fact. since the resultant product is the significant element and not the process of producingr it, the harmonic generation and the second order modulation could occur Yin a single modulatingr device by the lprocess sometimes called cascade modulation and this process and third order modulation could be used interchangeably.

The whole process of, single side band production detailed above may be generalized by stating the following rule which will be` found to apply to all cases. Two pairs of side bands must be produced which must have the same fixed frequency characteristic mp, where the side band frequencies are indicated by 'mp-Pq and 'mp-q, there must be a uniform relative phase shift of -2 between between the frequencies mp. When m be per- Leemans Since side bands of frequencies mptq and mp-q may be produced by modulation of the m+1 order using a carrier p, or by second order modulation using a carrier mp, or in several other ways, a generic expression is required, which is equally descriptive of these methods. Consequently, the words multiple and sub-multiple will be used in certain of the claims in their generic senses to include both the single and plural multiples or sub-multiples, unless such an interpretation would be inconsistent with the context.

Although the invention has been illustrated as embodied in a specific form, it should he understood that this is illustrative of only one of many possible forms in which the system of the invention may be embodied and that the scope of the invention is not to be limited by the particular form illustrated but only by the appended claims.

What is claimed is:

1. The method of producing a uniform relative phase shift of the band of frequencies in a multi-frequency current which couiprises transmitting 'individual portions of the current, each portion including all of the frequencies, by wave propagation to different electrical distances, whereby there tends to be produced a phase shift for .each frequency in each portion, which shifts are greater for the currents transmitted the greater distance and which increase with increase of frequency, and causing the increase in'shift with increase in frequency to be sufficiently less for the currents transmitted over the reater distance to produce an equal relative phase shift between the corresponding frequencies in the two portions.

2. The method of producing a uniform relative phase shift of the band of frequencies in a multi-frequency current which coniprises propagating individual portions of the current, each portion including all of the frequencies, by wave motion for different periods of time so as-to tend to produce phase differences between the corresponding frequencies in the two portions, the resultant phase difference being proportional to the frequency, and simultaneously compensating for this differential phase shift by causing the phase shifts of the frequencies in the portionvwhich is propagated the greatest length of time to increase less rapidly with increase in frequency than in the other portion.

3. A system for producing a uniform relative phase shift of a band of frequencies comprising in combination, a' source of frequencies constituting a band, two circuits branch from this source, a. delay circuit in each branch adapted to transmit each of the frequencies from said source, each said delay circuit comprising means for delaying the transmission of the frequency components in proportion to the frequencies in said ais lll() land superimposing a l of sai band, means whereby said delay circuits tend to produce different delays for each frefrequency, said difference increasing with the frequency and means affecting said delay-frequenc proportionality characteristic of said de ay circuits whereby the tendenc for said differences to increase with the flequency is compensated so as to.result irlilqtnniform difference, hence uniform phase s 1 4. The system of claim 3 in which the second means is adapted to produce a greater delay in one delay circuit than in the other and in which the last mentioned means is adapted to cause the delay to change more slowly with frequency increase in the delay circuit in which the, greatest delay occurs than in the other delay circuit.

5. The method which comprises dividing the current of a wave comprising a band of frequencies into two portionseach having all the frequency components of the wave and produclng any desired phase shift, in one portion with respect to the other portion, which is substantially constant for all frequencies within the band and conjointly utilizing the resultant Waves.

6. The method which comprises dividing a wave comprising a plurality of frequency components into a plurality of portions each containing all the frequency components and producin phase shifts of any desired amount, 1n some of the portions with respect to other portions, which are substantially constant for all of the plurality of frequency components and conjointly utilizing said components.

7. The method of producing a wave having the characteristics of a Single side band of a modulated wave which comprises generatin four modulated carrier side bands, two o? which have o posing hase relation,

side bands.

8. A system for producin a uniform relative phase shift of. a han of frequencies, comprising in combination a source of frequencies constituting a band, two circuits branched therefrom, a filter in each branch whose transmission range comprises each of the frequencies from said source and in which the phase change for frequencies with-V in this transmission range varies as a func-` tion of the frequency, said filters differing from each other by an odd number of sections, and the frequency transmission range of the lter having the greater number of sections being `sullinzeientlydgreaterY than thatof the other filter to pr uce a parallelism of their phase shift-frequency characterxstl curves. r f

9. The system of claim 8 inv whlch the phase change-frequency characteristic -curve of each of said"\1ters is substantial! g linear.

10.Thesystemofclaim8inw ychthe phase shift-frequency characteristic curve of each of said filters is substantially symmetrical about the point corresponding to the midfrequency of its transmission range.

11. The system of claim 8, in which the phase change-fre uency characteristic curve of each of said fi ters 1s substantially linear and in which the band occupies a` portion of the transmission range of each filter substantially in the center thereof.

12. The system of claim 8 in which the phase shift-frequency-` characteristic curve of each of said frequencies is substantially symmetrical about the point corresponding to the midffrequency of the transmission range and in which the band occupies a portion of the transmission range of each filter substantially at the center thereof.

13. The system of claim 8 in which the phase change-frequency characteristic curve of each of said filters is substantiall linear, is substantially symmetrical about the point corresponding to the mid-fre uency of its transmission range and in which the band occupies a portion of the transmission range of each filter substantially at the center thereof.

14. The method of producing a single side band carrier modulated wave which consists in dividing a modulating Wave having a plurality of frequencies g into two portions each having all of the frequencies of the Wave, producing a 90 uniform relative phase shift between the components of corresponding frequency in said portions, generating a single frequency wave of frequency p, changing said single frequency wave to an otherwise similar carrier wave which has two portions whose phases differ by an amount that would correspond to a 90 phase difference between the same portions if each were converted to a wave havin a frequency mp, utilizing a portion of sai single frequency wave and one portion of the modulating wave to produce side bands of frequencies mp+ and mp-q in which m is an lnteger, including intermodulating the portion of the modulatlng wave with a carrier wave whose frequency is sub-multiple of the frequency mp, utilizing another portion of said single fre uency wave and the other portion of the moculating wave to produce side bands of frequencies mp-I-g andV mpg, and superposmg the resultant pairs of side bands.

15; `The method of producing a single side band carrier modulated wave which consists of dividing a modulating wave having a lurality of frequencies g into two portions each having` all of the frequencies of the wave, producing a 90 uniform relatiyey phase s ift between the frequencles 1n said portions .generatin a carnet frequency wave,`div1din sai carrier wave luto two portions, p ucing a 90 relative phase shift between the phases of said carrier wave portions, interluodulating each portion of said carrier wave with a portion of' the modulating Wave to produce side bands of said carrier wave, and superposing the resultant pairs of side bands whereby one side band of each pair is balanced out and the remaining side bands are added.

16. A system for producing a single side band carrier modulated wave, which comprises in combination a source of multi-frequency modulating waves, two circuits branched therefrom, means in said branched circuits for producing a uniform relative phase shift of frequencies in the respective branch circuits, a source ot' carrier frequency, means for deriving from said source two carrier Waves each having a fre uency of the Wave from said source but di ering from each other in phase by 90, means for modulating each of said derived carrier waves with the output wave from each of said branch circuits to produce upper and lower side bands of each of said derived carrier Waves, and means for superposing the resultant pairs of side bands, whereby one side band of each pair is balanced out and the remaining side bands are added.

17. The system of claim 16 in which the means for producing a uniform relative phase shift of the modulating frequency comprises a filter in each of said branch circuits, said filters dii'ering from each other by an odd number of sections, the filter having the greater number of sections also having a siliciently greater transmission range to compensate for the differential phase CERTIFICATE Patent No. 1,666, 206.

RALPH V.

It is hereby certified that error appears in the printed specification of the shift-frequency characteristic resulting from the use of a. different number of sections.

18. The method of producing a composite signaling wave comprising producing two separate composite Wave trains having at least two frequency components in common, producing a phase shift 1n at least one wave train which changes the phase relation between the corresponding components of the two trains, the phase shift difference between correspondlng components of one fre- (fluency being the same as that for the other requency or frequencies, and conjointly utilizing said wave trains.

19. A method of roducing a single sideband carrier modu ated wave which comprises generating a modulating frequency band, coverting said band into two portions which have the same fre uency characteristics but in which there isqbetween the comonents of the same frequencies occurring 1n the respective portions, a phase difference which is the same for all components, and utilizing such portions.

20. In combination, means for generating a band of frequencies, means for dividing said band of frequencies into two portions each of which includes all of the frequencies, means adapted to operate on said portions to cause any desired uniform relative phase displacement between correspondng frequency components thereof, and means to jointly utilize the resultant ortions.

In witness whereo I hereunto subscribe my name this 14th day of January A. D.,

1925. RALPH V. L. HARTLEY.

oF CORRECTION.

cmad April i1, ma, to

L. HARTLEY.

above numbered patent requiring correction as follows: Page l, line 22, after the word l'been" insert the word "to"; page 2, lines 67 and 68, italicize the word "relative"; page-3, line 127, strike sert a plus sign; page 4, in the equa "n sub 2" read "b sub 2"; page 5, line 100, for the word "is linea 52 and 53 for the misspelled word "parrallel" read "parallel" "parallelism'h page 7, line 23, for the word "would" read e "ire"; and that said Letters Patent ha'tthe same may conform to the record for "perellelisin" read "conld"; page 9, line 2, strike ont the sylrlabl should be read with these corrections therein ,t

oi the case in the Patent Office.

Signed and sealed this llth d ay (Saal) out the division sign in the exponent and intion following line 106, for the denominator of Septeniber, A. D. 1928.

M.J.Monre,

Actin Comisiones' of Patents.

u read Nitti; page 6. land line 55,

shift between the phases of said carrier wave portions, interluodulating each portion of said carrier wave with a portion of' the modulating Wave to produce side bands of said carrier wave, and superposing the resultant pairs of side bands whereby one side band of each pair is balanced out and the remaining side bands are added.

16. A system for producing a single side band carrier modulated wave, which comprises in combination a source of multi-frequency modulating waves, two circuits branched therefrom, means in said branched circuits for producing a uniform relative phase shift of frequencies in the respective branch circuits, a source ot' carrier frequency, means for deriving from said source two carrier Waves each having a fre uency of the Wave from said source but di ering from each other in phase by 90, means for modulating each of said derived carrier waves with the output wave from each of said branch circuits to produce upper and lower side bands of each of said derived carrier Waves, and means for superposing the resultant pairs of side bands, whereby one side band of each pair is balanced out and the remaining side bands are added.

17. The system of claim 16 in which the means for producing a uniform relative phase shift of the modulating frequency comprises a filter in each of said branch circuits, said filters dii'ering from each other by an odd number of sections, the filter having the greater number of sections also having a siliciently greater transmission range to compensate for the differential phase CERTIFICATE Patent No. 1,666, 206.

RALPH V.

It is hereby certified that error appears in the printed specification of the shift-frequency characteristic resulting from the use of a. different number of sections.

18. The method of producing a composite signaling wave comprising producing two separate composite Wave trains having at least two frequency components in common, producing a phase shift 1n at least one wave train which changes the phase relation between the corresponding components of the two trains, the phase shift difference between correspondlng components of one fre- (fluency being the same as that for the other requency or frequencies, and conjointly utilizing said wave trains.

19. A method of roducing a single sideband carrier modu ated wave which comprises generating a modulating frequency band, coverting said band into two portions which have the same fre uency characteristics but in which there isqbetween the comonents of the same frequencies occurring 1n the respective portions, a phase difference which is the same for all components, and utilizing such portions.

20. In combination, means for generating a band of frequencies, means for dividing said band of frequencies into two portions each of which includes all of the frequencies, means adapted to operate on said portions to cause any desired uniform relative phase displacement between correspondng frequency components thereof, and means to jointly utilize the resultant ortions.

In witness whereo I hereunto subscribe my name this 14th day of January A. D.,

1925. RALPH V. L. HARTLEY.

oF CORRECTION.

cmad April i1, ma, to

L. HARTLEY.

above numbered patent requiring correction as follows: Page l, line 22, after the word l'been" insert the word "to"; page 2, lines 67 and 68, italicize the word "relative"; page-3, line 127, strike sert a plus sign; page 4, in the equa "n sub 2" read "b sub 2"; page 5, line 100, for the word "is linea 52 and 53 for the misspelled word "parrallel" read "parallel" "parallelism'h page 7, line 23, for the word "would" read e "ire"; and that said Letters Patent ha'tthe same may conform to the record for "perellelisin" read "conld"; page 9, line 2, strike ont the sylrlabl should be read with these corrections therein ,t

oi the case in the Patent Office.

Signed and sealed this llth d ay (Saal) out the division sign in the exponent and intion following line 106, for the denominator of Septeniber, A. D. 1928.

M.J.Monre,

Actin Comisiones' of Patents.

u read Nitti; page 6. land line 55,

Images