US2545871A - Apparatus for compressing or expanding the frequency bands of electric oscillations - Google Patents

Description

March 20, 1951 Filed June 5, 1948 D. A. BELL APPARATUS FOR COMPRESSING OR EXPANDING THE FREQUENCY BANDS OF' ELECTRIC OSCILLATIONS 2 Sheets-Sheet 1 March 20, 1951 D. A. BELL 2,545,871

APPARATUS FOR COMPRESSING OR EXPANDING THE Patented Mar. 20,v 1.951

APPARATUS FOR A(DOMPRESSING R EX- PANDING THE FREQUENCY BANDS ELECTRIC OSCILLATIONS David Arthur Bell, Taplow, England, assigner to British Telecommunications Research Limited, Taplow, England, a company of Great Britain n Application June 5, 1948, Serial No. 31,294 In Great Britain June 5, 1947 The present invention relates to apparatus for compressing or expanding the frequency bands of electric oscillations. Frequency band compression and expansion will be referred to hereinafter for shortness as frequency transformation.

It is known that the signal/noise ratio in an electrical communication system is a function of band-width, and that an improvement in signal/noise ratio may be obtained in frequency modulation systems by Aincreasing the overall frequency band occupied during the transmission of a given signal, and in amplitude-modulation systems by reducing the maximum frequencies contained in the given signal. The latter causes a reduction of quality or distortion of the signal, but this may sometimes be acceptable in return for either a reduction in transmission bandwidth or an improvement in signal/noise ratio. The theoretical background of this kind of operation is described in a paper in the March 1948 issue of Electronics page 138, under the title Theoretical Requirements for Communication Systems. One of the systems used for such transformations of frequency band is that known as pulse code modulation, but there is an alternative method in which the transformation is elfected by providing means which directly reduce all frequencies in the signal in the same proportion.

In a paper in The Journal of the Institution of Electrical Engineers vol. 93 part IllNo. 26 pages 445 to 454, there is described apparatus for transforming the frequencies of bands of electric oscillations, in which a recording, known as a sound track, of oscillations whose frequency band is to be compressed is produced on a film. The film is then moved past a reproducing device including a number of slits which are given a com. ponent of motion in the same direction as the di, rection of movement of the film, but a different velocity less than that of the lm, or more than, but less than twice that of the film. The band of frequencies occupied by the oscillations reproduced by the reproducing device is then compressed, in comparison with the frequency band of oscillations which would be derived by a stationary slit, by an amount dependent upon the ratio of the velocities of the slit and the fllm.

It should be pointed out that in the transformation with which the above-mentioned paper, and the present invention, are concerned, the frequency band occupied by oscillations representative of intelligence is increased or decreased whilst the time occupied in transmission of the intelligence .snot altered. It is thus dis- 5 Claims. (Cl. 178-44) tinguished from the result obtained by merely running the nlm past a stationary lreproducing slit at an increased or decreased speed. It is essential that the width of the slits used should be I 'suiiiciently small in relation to the wavelength,

upon the film, of the highest frequency in the band to be transformed.

In the arrangement described in the paper above referred to, a length of film dependent upon the size of a window is in position to cooperate with one or more slits at any one time. The window must not be so long in the direction of travel of the film that a slit operates on a section of the film` which is intended to be transmitted at an appreciably different time. After traversing the window a slit goes out of action and it is therefore arranged that other slits are brought into the window in succession. In order to maintain continuity of transmission the minimum requirement is that as one slit is leaving the windowfa succeeding slit is `just enonly a portion of the reproduced signal is terminated as each slit leaves the window. Furthermore the transparency of the window may be graded from a, maximum at the middle thereof to zero at lits sides. Thus each slit commences and ends its action gradually instead of abruptly.

On pages 454 and 455 of the same paper it is suggested that frequency transformation may be effected by electrical means.

It is the principal object of the present invention to provide novel apparatus for transforming the frequencies of electric oscillations by purely electrical means, whereby the transformation can bev effected without the use of moving parts.

According to the present invention, apparatus for transforming the frequencies of electric oscillations comprises two delay networks having different delay times and substantially non-reflectiVe terminations, means for applying oscillations over a band of frequencies, which band is to be transformed, to the input of one delay network,

means for applying voltage pulses to the input lof the other network, and an output circuit y coupled to the delay networks at a plurality of delay times from the inputs of the two networks being substantially the same, the output circuit being arranged in such a manner that output oscillations reproduced therein are representative of the applied oscillations, but lie in a band whose width is dependent upon the said ratio.

According to a feature of the invention the delay networks comprise an equal number of sections connected in cascade respectively, each section containing lumped impedance elements and the sections of each network being like, and the output circuit comprises a plurality of multiplying devices, said pairs of points being constituted by the junctions of correspondingly numbered sections of the two networks, counting from their input ends, and said multiplying devices having input terminals coupled respectively to the two networks at the pairs of points Aand having output circuits coupled to a common output terminal.

According to a further feature of the invention the delay networks have inductance and capacity distributed along the lengths thereof and the output circuit comprises an element of material having non-ohmic conductivity in contact with the delay networks along the lengths thereof and having associated therewith electrodes between which an output voltage is derived.

In this case the number of pairs of points may be very large.

Other features of the invention will be apparent from two examples of apparatus for carrying the invention into eifect which will now be described with reference to the accompanying drawing, in which Fig. 1 is a schematic diagram of one embodiment of the invention wherein the two delay networks each comprise a plurality of sections connected in cascade, each section consisting of lumped impedance elements, Fig. 2 is a diagram of a multiplying device shown in blocked form in Fig. l, Fig. 3 is a schematic diagram of a second embodiment of the invention in which the two delay networks have their inductance and capacity distributed along the lengths of the networks, and Figs. 4 and 5 illustrate one practical arrangement of part of the embodiment shownin Fig. 3. All parts performing like functions are given like references in all gures.

Referring to Fig. l two delay networks 3 and 4 are arranged to have different delay times and are terminated by substantially non-reflective terminations 8 and 5 respectively. The delay network 3 comprises a number of like sections I4 of which four are shown each including lumped inductance I5 and capacitance I6. The network 4 comprises an equal number of like sections I1 each section including lumped inductance I8 and capacitance I9.

A source I of oscillations occupying a frequency band to be transformed is connected to .the input end of the network 3. A pulse generator 2 is connected to the input of the network 4 for applying thereto voltage pulses whose duration (or width) is preferably small compared with the period of the highest frequency in the band to be transformed.

Each section I4 of the network 3 provides an equal time delay for oscillations passing through the network, each section I'I provides an equal delay for pulses passing through the network 4, and the time delay provided by each section I4 is arranged to diifer from that provided by each section I'I.

Multiplying devices 'I are connected as shown by means of connecting leads such as I0 and 9 respectively, between the networks 3 and 4 at their input ends, between successive junctions of correspondingly numbered sections of the networks counting from the input, and between the networks at their output ends. rIhese devices have a common output lead 8 and will be described later.

Thus at each of the points in the delay networks `at which 'the .multiplying devices 'I are connected thereto, the ratio of the delay times from the input ends of the networks is the same.

Turning now to Fig. 2, this shows a suitable type of multiplying device 'I (Fig. 1). A thermionic valve 2D has one control electrode II connected to receive pulses from the network 4 (Fig. 1) through 'a condenser 2I and the lead 9. Oscillations from the network 3 are applied to another control electrode I2 through the connecting lead I0, an attenuator shown between two terminals 29 and 'I3 and a condenser 22 and grid leak 23. The function of the attenuator will be described later. The valve 2G is biased beyond anode current cut-oi'f by a battery 24 in order to render the valve inoperative except on the application to the grid I I of a pulse which in this embodiment is assumed to be positive. The control electrode I2 is biased by the same battery 24 for normal operation when the valve is rendered operative by a positive pulse on its grid II.

Thus the valve 20 acts as a gate to the oscillations applied to the control electrode I2, and is open only when a gating pulse is applied to the electrode II.

The output vcircuit of the valve 20 comprises a load resistor `25 across which output oscillations are developed, and the l.anode of 'each valve 2B in the devices 'I (Fig. 1) is connected to the common output lead .8.

In order to assist in the understanding of the action of this embodiment, it 'will be assumed first of all that a single sinusoidal oscillation is applied to the input of the network 3 Vat the instant when the instantaneous amplitude :of 'the oscillation is zero. It will :also be assumed that one pulse is applied at the same instant to the input of the network 4, and that contrary to the actual case described) the time delays of the two networks are equal.

Inlthis case as the pulse 'renders each consecutive device 'I operative, the instantaneous value of the oscillation applied to -the device remains at zero and the frequency of the oscillation present inthe output lead Bis zero.

Assuming new that the time Ydelay Yof the network 4 lis increased vby a small amount, the instantaneous values of the oscillation applied -to the'device at the instants of application of pulses thereto, changes from `device to device by an amount dependent upon the ratio -of the time delays. Hence the output present inthe -line 18 is in the -form of pulses whose amplitude var-ies at a frequency fo given `by f0=fi\(t1-t2)/t1 where fi is the frequency of the oscillation applied to the network 3, t1 is the ltime delay provided `by the network 3, and t2 is the time Adelay of the network 4.

Hence it wil-l be appreciated 4that transformation of the frequencies of a band of oscillations applied to the network 3-can be effected with the arrangement described. By arranging .t1 to be smaller than t2 frequency band compression 4can be .carried out and frequency `band expansioncan be effected .by making tl larger thantz.

f: Itis desirable that only asmall number of 'pulsesshould be present in the delay network at any one instant. This can .be arranged by suitable selection of the recurrence frequency of the pulse generator 2 (Fig. 1) in relation to the total delay time of the network 4. Furthermore it is desirable to ensure that the4 only one of the devices l' should contribute amaximum to the output oscillations at any one instant. Thus taking one pulse traversing the delay line 4, the contribution to the output oscillations caused by the pulse rendering successive devices 'I operative, should commence at a low value at the input end of the network, increase to a maximum at about the centre of the network and fall to a low Value at its termination.

This is provided for in the arrangement shown in Fig. 2 by means of the attenuator connected between theterminals 29 and I3. It is arranged that the degree of attenuation is high in the device 7 connected at the input and output ends of the networks S and d andV that the degree of attenuation falls gradually to a minimum in the control device or devices.

The desired degrees of attenuation may be ef-` fected in other ways, for example by providing different values of Vload resistors 2E or by suitably grading the anode Yvoltages supplied to the valves 23iin the devices l. The grading of the attenuation may follow a variety of different laws, for instance one of those described in the paper above referred to for grading the transparency of the window.

It will be appreciated that this arrangement is somewhat analogous to the arrangementv described in the aforementioned paper.

The moving record of the oscillation is simulated by the wave passing along the delay network t, the moving slits by the pulses moving along the network 4, and the length of the window by the length of the delay line 3. In the embodiment described however the pulses are only utilised at intervals along the delay network. This is somewhat analogous to the use of a slotted window. The grading of the window is simulated by the arrangement of attenuators associated with the devices 1.

A closer analogy to the arrangement described in the paper4 referred to, is shown in Fig. 3. The delay networks 3 and in this embodiment comprise continuously wound inductors I5 and I8 and distributed capacity I6 and I9 respectively. The two delay networks are in contact along their lengths with an element 26 of material having non-ohmic conductivity. An example of suitable material is silicon carbide in the form known commercially as Atmite Thus the element 26 has a non-linear voltage/current characteristic and hence the current flowing through it isa function of the potentials on the two delay networks. The current thus includes components proportional to the products of the applied oscillations and pulses, and as the phase or time relation between a pulse and the applied oscillations varies as the pulse traverses the delay network 4, by an amount dependent upon the ratio of the time delays of the networks 3 and 4, the output oscillations include oscillations having frequencies of the previously derived form rIwo electrodes 21 and 28 are inserted in, or are arranged on, the element 26 in such a manner that they do not interrupt the circuit existing in the element 2S between the two networks.

The currents owing in the e1ement=26 develop potential differences across these electrodes 21 and 28 which are connected to output terminals B and respectively.

In this embodiment the grading effect may be produced by suitable variation of the composition or cross-section of the element 26.

In designing delay networks for use at audio frequencies it; may be found convenient to wind the inductors I5 and I8 on cores of magnetic material in order to obtain suiiicient inductance per unit length. Furthermore in order to provide sui'ncient capacitance IS and I9, particularly when the inductances are air-cored, it may be desirable to wind the inductors I5 and I8 on a former or" conducting material.

Referring to Fig. 4, this is an end view of one practical form which the coils I5, I8 and element 26 of the embodiment of Fig. 3 maytake. The inductors I5 and I8 are wound on cores 30 and 3| respectively of magnetic material and square cross-section. The element 26 is disposed between the two inductors I5 and I6 and in contact with every turn thereof.

Fig. 5 is an exploded View of the arrangementv of Fig. 4 and shows that the element 26 has its longitudinal cross-section tapered from a maximum in the centre to a minimum at both ends. The cross-section is taperedv in this way in order to effect the aforesaid grading.

It will-be appreciated that a suitable filter must be connected to the output terminals B and 8 to select the desired frequency band and reject undesired frequencies.

1. Apparatus for transforming the frequencie of electric oscillations, the apparatus comprising two delay networks having different delay times and substantially non-reflective terminations, said delay networks comprising an equal number of sections, each said section comprising lumped impedances and said sections of each said network being substantially identical with one another, means for applying to the input of one of said networks oscillations over a band of frequencies, which band is to be transformed, means for applying pulses to the input of the other of said networks, a plurality of pairs of tapping points in said networks, the points in each said pair lying in said two networks respectively, and the ratio of the delay times from the inputs of said networks to the tapping points in all said pairs being substantially the same but other than unity, and multiplying means couplinga common output terminal to said two delay networks at the junctions of correspondingly numbered sections of said networks, counting from their input ends.

2. Apparatus for transforming the frequencies of electric oscillations, the apparatus comprising two delay networks having different delay times and substantially non-reective terminations, said delay networks comprising an equal number of sections, each said section comprising lumped impedances and saidv sections of each said network being substantially identical with one another, means for applying to the input of one of said networks oscillations over a band of frequencies, which band is to be transformed, means for applying pulses to the input or the other of said networks, a plurality of pairs of tapping points in said networks, the points in each said pair lying in said two networks respectively, and the ratio of the delay times from the inputs of said networks to the tapping points in all said pairs being substantially the same 7 but :other than unity, and a vcommon output circuit 'comprising a plurality of multiplying devices coupled respectively to said two networks at the junctions of correspondingly numbered sections of said networks, counting from their input ends.

3. Apparatus for transforming the frequencies of electric oscillations, the apparatus comprising two delay networks having different delay times and substantially non-reliective terminations, means for applying to the input of one of said networks oscillations over a band of frequencies, which band is to be transformed, means .for applying .pulses to the input of the other of said networks, 'a plurality of pairs of tapping points in said networks, the points in each said pair lying in said two neworks respectively, and the ratio of the delay times from the inputs of said networks to the tapping points in all said pairs being substantially the same but other than unity, and a common output circuit coupled to said plurality of pairs of tapping points, said output circuit comprising a plurality o1 multiplying devices, means for coupling said multiplying devices to said pairs of tapping points respectively, a common output terminal and means for coupling the multiplying devices to said common output terminal.

4. Apparatus for transforming the frequencies of electric oscillations, the apparatus comprising two delay networks having different delay times and substantially non-reflective terminations, means for applying to the input of one of said networks oscillations over a band of frequencies, which band is to be transformed, means for applying pulses to the input of the other of said networks, a plurality of pairs of tapping points in said networks, the points in each said pair lying in said two networks respectively, and :the ratio of the delay times from the inputs of vSaid networks to the tapping points in all said pairs being substantially the same but Aother than unity, andan output circuit coupled to said plu rality vof pairs of tapping points, said output circuit comprising `a plurality of multiplying devices, vmeans .for coupling said Ymultiplying devices to .said pairs rof tapping 'points respectively, a common output terminal and means for coupling the .multiplying devices to Asaid common output terminal, each said multiplying device comprising an electron discharge A,device `having an anode, a cathode Vand at Aleast two control electrodes, and said means vvfor coupling said multiplying devices to .said pairs of .tapping points comprising means for coupling said control electrodes of 'said `discharge devices to said pairs of tapping points.

5. Apparatus as claimed in claim 4, wherein means are provided :for rendering said discharge device in each of said multiplying devices substantially non-conductive in the absence of pulses applied to said discharge vdevice from said other ofthe networks.

DAVID ARTHUR BELL.

REFERENCES CITED The following references are of record in the ille of this patent.:

UNITED STATES PATENTS Number Name Date 1,671,143 Campbell May 29, 1928 2,172,354 Blumlein Sept. 12, 1939 2,410,233 Percival v-- Oct. 29, 1946

Images