US3020352A

US3020352A - Single sideband equipment for speech transmission

Info

Publication number: US3020352A
Application number: US646237A
Authority: US
Inventors: Jager Frank De; Greefkes Johannes Anton
Original assignee: US Philips Corp
Current assignee: US Philips Corp; North American Philips Co Inc
Priority date: 1956-04-26
Filing date: 1957-03-15
Publication date: 1962-02-06
Anticipated expiration: 1979-02-06
Also published as: GB847626A; DE1032345B; BE557017A; ES235000A1; NL102595C

Description

Feb. 6, 1962 F. DE JAGER ETAL 3,020,352

SINGLE SIDEBAND EQUIPMENT FOR SPEECHTRANSMISSION Filed March 15, 1957 4 Sheets-Sheet 1 T if K ...l l $111# Feb. 6, 1962 F. DE JAGER ETAL 3,020,352

SINGLE SIDEBAND EQUIPMENT FOR SPEECH TRANSMISSION Filed March l5, 1957 4 Sheets-Sheet 2 t l k lcluo lio rizo l` f' l l l I l l I I m m, zo 4o 6o eo loo |20 |40 g I l o Fl G2 INVENTo FRANK DE. JAGER JOHANNES ANTON GREEFKES Feb. 6, 1962 F. DE JAGER EVAL SINGLE SIDEBAND EQUIPMENT FOR SPEECH TRANSMISSION Filed March 15, 1957 "SUQGLE,

SHDEBAND FIL-TED` 4 Sheets-Sheet 3 OSCI l-I-ATO Q INVENTOR FRANK DE JAGER JOHANNES ANTON GREEFKES AGENT Feb. 6, 1962 F. DE JAGER ETAL 3,020,352

SINGLE SIDEBAND EQUIPMENT FR SPEECH TRANSMISSION Filed March 15, 1957 4 Sheets-Sheet 4 osmu. kroq l ,klNvENToR 4 In FRANK DE JAGER J OHANNE S ANTGJ BY GREE FKES` AGENT United States This invention relates to single sideband equipment comprising a transmitter and a receiver for the transmission of speech signals, the transmitter comprising a single sideband generator controlled by speech signals and a limiter for limiting the single sideband speech signals to a substantially constant amplitude, which limited single sideband signal is transmitted to a receiver comprising a speech signal cascade of circuits in which the incoming single sideband signal is supplied through a single sideband filter and a limiter to a single sideband demodulator and a receiver output connected thereto. The invention also relates to transmitters and receivers for use in the described equipment.

Such a single sideband equipment, which was described in applicants prior application Serial No. 542,305, led October 24, 1955, now Patent No. 2,907,831, achieves a very high transmission effect since in contradistinction with conventional single sideband equipment, now on the one hand the transmitter normally can be fully loaded by the limited single sideband signal and, on the other hand, readily usable reception of speech is still possible at comparatively low signaletonoise ratios at the receiver input, for example l5 db and less.

By furthermore transmitting, in accordance with the said prior application (for example by means of narrowband frequency modulation) a unipolar envelope voltage proportional to the instantaneous intensity of the single sideband speech signals and containing frequencies up to at most the lowest of the speech frequencies to be transmitted transmission of speech has been found possible even at signal-to-noise ratios of about 12 to 9 db at the receiver input.

An object of the invention is, in single sideband equipment of the kind described in the preamble, to obtain a greatly decreased influence of noise and other interference upon the speech transmission and to render speech transmission possible at exceptionally low signal-to-noise ratios at the receiver input, for example about 3 db.

According to the invention, for this purpose the transmitter comprises means for producing and transmitting to the receiver a switching voltage which varies between fixed values in accordance with the instantaneous presence or absence of letter sounds, whilst the receiver comprises means for receiving the switching voltage, and the speech signal cascade of circuits is provided with a blocking switch controlled by the incoming switching voltage, for blocking or opening the speech signal cascade in the instantaneous absence or instantaneous presence of letter sounds.

When using the invention, in contradistinction with said prior application, there is not transmitted an envelope voltage, the amplitude of which is proportional to the instantaneous intensity of the speech signal. Instead of such a continuously variable envelope voltage, a switching voltage varying between fixed values is now transmitted and hence a signal having the character of a telegraph signal. Use is preferably made of a switching voltage which varies between only two fixed values (for example 0 and 1), in other words a bivalent switching voltage, although it is fundamentally possible to utilize a trivalent or a multivalent switching voltage. Since, now, a switching voltage varying between fixed values tit is transmitted, the inuence of noise voltages and interfering signals upon the transmitted and comparatively narrowband switching voltage may be substantially eliminated at the receiving end by means of several regenerating means known per se for signals of a telegraph character, for example bilateral limitation (slicing), bistable trigger circuits, etc.

The described method of transmitting speech signals with separate transmission of, on the one hand, the frequency at normal bandwidth by means of a limited single sideband signal and, on the other hand, an amplitude in code form which by its nature requires a comparatively narrow frequency band only, will hereinafter be briefly referred to as Frenac.

A switching voltage indicating the instantaneous presence or absence of lettersounds may be produced in greatly different ways. In analogy with the description in the priorapplication, it is possible to start from the single sideband speech signal, which may be supplied to an amplitude detector for obtaining a unipolar envelope` voltage, followed by means for converting the envelope' voltage into a switching voltage having a value 0 in the instantaneous absence of an envelope voltage and a valuev possible of the initial single sideband speech signal with continuously variable amplitude of the envelope.

In accordance with the present invention, the object of the transmission of the switching voltage is only to provide an indication of the instantaneous presence or absence of letter sounds. In connection therewith it is frequently more advantageous to derive this indicationv directly from the low-frequency speech signal instead of, as previously mentioned, indirectly from the single sideyband speech signal, in which latter case, for example,

carrier-wave leakage of thetsingle sideband modulator and low-frequency interference may have a disturbing influence by masking of a comparatively low envelope voltage.

The switching voltage which is similar to a telegraph signal may naturally be transferred in different ways known for telegraph transmission (for example A1, A2, F 1). For example, it is possible to utilize a pilot signal, the amplitude of which has a value zero or a given finite value dependent upon the instantaneous value of the switching voltage. The frequency of the suppressed carrier wave for the transmitted single sideband signal may be chosen as the pilot frequency.

In order to achieve a constant load on the transmitter,

the transmitter is preferably so designed that the pilot` voltage and the limited single sideband signal do not occur simultaneously, but voccur alternately and both load the transmitter substantially fully.

The switching voltage may also be transferred in a manner as usual for frequency shift telegraphy, that is t0 say by `transmission of a pilot signal, the frequency of which shifts to and fro in the rhythm of the switching voltage between predetermined fixed frequency values.

If desired, use may be made of frequency-diversity circuits for transmission of the switching voltage, as is also t known per se for telegraph signals, for the purpose of des creasing the influence of, for example, selective fading .in` The influence of noise upon the switching voltage in thisvcase the radio transmission. of the switching voltage.

will usually be greater.

The switching voltage transmitted in accordance with the invention has the function to cut off the speech signal cascade of circuits in the receiver and thus to prevent the transmission of noise to the receiver input during inter alia small intervals between letter sounds. Preferagbly, such cutting off must already be etfected at very small intervals of about 0.03 sec. such as occur between the letter sounds n and ch in the word benchl (cf. in this connection Speech and Hearing in Communication, by H, Fletcher, editor, D. Van Nostrand Company, Inc., New York, 1953, page 26, FIG. 2l). Otherwise it is advantageous even for the shortest consonants asfpi and t, which represent a duration of 0.02 to 0.0i sec. (cf. again FIG. 2l) to be transmitted almost cornpletely to the receiver output, thus preferably permitting such a transient to open the receiver cascade of circuits. Taking the foregoing into consideration, it is advantageous if the maximum switching voltage frequency that may be transferred. is higher than 30 c./s. and is, yfor example, from 0 to 100 c./s.

When the maximum switching frequency that may be transferred is reduced to, for example, about c./s., a distinctly noticeable decline in quality of the transmitted speech occurs, which is attributable inter alia to unwanted shifts of the times at which the speech signal cascade of circuits is opened for the transmission of explosivel consonants or cut o for the suppression of noise between letter sounds.

When the maximum switching frequency which may be transmitted is chosen to be higher than 100 c./s., for example 150 c./s., a reproduction of explosive sounds occurs which is more accurate in time, whilst under certain conditions the receiver cascade is switched on and off in the rhythm` of the fundamental tone frequency of vowels. The improvement in quality achieved by increasing the maximum transferable switching voltage frequency above 1,00 c./s. has been found to be comparatively small since, for example,v the periodicity of the fundamental tone frequency is also represented already in the limited single sideband signal due toy interference of speech frequency components. Only the improved suppression of noise by blocking and opening the receiver cascade at exactly the correct moments does appear to add to the improved quality of the transmission, but the worse signal-to-noise ratio counteracts-said improvement in the quality of the transmission due to the required greater band-width for the' switching voltage. An increase of the maximum transferable switching voltage frequency to the lowest transmitted speechy frequencies in the single sideband signal, for example about 200 to 350 c./s., is thus of little or no consequence, at least for signal-to-noise ratios which on the average are low.

A maximum transferable switching voltage frequency of from 50 to 10,0 c./s-. has experimentally been found to be optimum for the. transmission of speech signals in the Dutch language at low signal-to-noise ratios, simulated by admixture of (White) noise in the transmission path between the. transmitter and the receiver.

Substantially normal usable speech transmission and reasonable recognizability of voices has been found -to be possible already at a signal-to-noise ratio of only 3 db for the transmitted single sideband having a band-width of about 3000 c./s. with the use of a pilot signal for the transmission of the switching voltage which is present alternately with letter sounds,

At signal-to-noise ratios which are only higher (5 db and higher), a speech transmission showed excellent intelligibility and recognizability of voices, especially with the use of means at the receiving end, which will be discussed hereinafter, for rounding 0E the regenerated switching voltage.

At a signal-to-noise ratio lower than 3 db, for example 0.*db, it appeared to be fairly well possible to follow the speech, provided thatthe words were spoken slowly and clearly.y articulated. Upon further decrease of the signalto-noise ratio and hence at signal-to-ratios below 0 db, the practical usability of the speech communication rapidly decrease, inter alia due .to the signal-to-noise ratio becoming too unfavourable for a switching voltage transmitted by amplitude modulation.

As previously mentioned, the switching voltage serves inter alia to open and block the speech `signal cascade in the receiver upon occurrence of transient explosive consonants and occurrence of small intervals between sequential letter sounds. With the choice of the maximum transferable switching voltage frequency it is necessary to make allowance -for the speaking velocityusual for the language to be transmitted that is to say the usual speed of movement of the sound-producing talking organs, such as mouth cavity, lips, etc. The talking speed, for example, for French and Italian, is a little higher than for English and Dutch. in view thereof, i-t is preferable for equipment intended for universal use to chose the maximum transmitted switching voltage frequency equal to approximately c./s.

in order that the invention may be readily carried into effect, it will now be described more fully, by way of example, with reference to the accompanying drawings in which:

FIGS. la and lb each show a block diagram of one embodiment of a Frenac transmitter and a Frenac receiver.

FIG. 2 shows voltage-time diagrams which serve to explain the production and use at the receiving end of the switching voltage used in accordance with the invention.

FIGS. 3a and 3b show detail diagrams of those parts of a Frenac transmitter and a Frenac receiver, respectively, which are essential for correct understanding of the invention.

In the block diagram of a Frenac transmitter as shown in PIG. la, reference numeral 1 indicates a microphone, from which the speech signals to be transmitted are derived. In view of the limitation of the single sideband speech signals which is to be effected in the transmitter and which-affects the high frequencies in the speech signal which are usually weaker, the speech signals derived from microphone l* are led via apre-emphasis network 2. The network 2 benets speech frequencies above, for example, 600 to 800 c./s., that is to say benefits higher speech frequencies by, for example, 6 db per octave. The speech signals are supplied through an amplifier 3 to a single sideband modulator 4, which has also' applied to it a carrier-wave frequency fo provided. by a carrierwave generator 5. The output circuit of the single sideband modulator 4 includes a single sideband filter 6', which passes the upper sideband produced upon modulation and which suppresses the carrierwave fo, together with the lower sideband. The pass-band of the single sideband filter 6 at a carrier-wave frequency fo of 60 kc./s. may be, for example, from 60.3 to 63 kc./s.

The single sidebandV signal derived from the single sideband lter 6 is moderately limited in a first limiter 7` and subsequently supplied through a make contact 8 of a relay 9 to a second limiter 10. In the output circuit of the second limiter 10, the single sideband speech signals are drastically limited, so that the sum of the amplitudes, of the different sideband components is limited to a coustant amplitude and the information carried by the signal is thus retained only in the frequency components of the single sideband signal. Distortion products located outside the single sideband to be transmited are produced as a result of this limitation. To suppress these distortion products, the second limiter 10 if followed by another single sideband lter 1,1, which for reasons to be discussed hereinafter, passes not only the said upper sideband, but also the carrier-wave frequency and hence, for example, has a passband offrom 59.9 to 63 kc./s.

The output of low-frequency amplifier 3 has also connected to it a switching voltage channel for producing a switching voltage indicating the instantaneous Vpresenceor absence of letter sounds from the speech signal. This channel includes an amplitude detector 12 for detecting the envelope oi the speech signal transmitted. To prevent hum voltages of, s-ay, 50 or 100 c./s. and other interfering signals adjacent in frequency from adding to the wave form of the envelope, the amplitude detector 12 is preceded by a lter 13, which passes only frequencies between 0.5 and 3 kc./s. without attenuation. The envelope voltage derived from amplitude detector 12 is smoothed by means of -a low-pass filter 14 having a cut olf frequency of from 100 to 200 c./s. The smoothed envelope voltage, which is unipolar, governs, after being amplified in a direct-current ampliiier 15, the above-mentioned relay 9, having the work contact 8 and a back contact 16. The rel-ay 9 serves to convert the envelope voltage with a continuously variable amplitude into a bivalent switching voltage. In View of the maximum value of the envelope voltage, the response sensitivity of the relay 9 must be chosen to be such that response already occurs as soon as the instantaneous value of the envelope voltage attains a threshold value of several percent, for example 2% to 4%, of the maximum instantaneous value of the envelope voltage. Otherwise, the relay 9 must of course be of the quick-switching type having a response time of, for example l msec., on the one hand for properly following switching voltage frequencies of from 100 to 200 c./s. and, on t-he other hand, to prevent time shifts between the moments when the envelope voltage exceeds the said threshold value of, for example, 2% to 4% and the moments when the change-over contact of the relay is changed from back contact 16 to make contact 8. It has been found from experiments that time shifts between said moments of about l to 2 msecs. are permissible.

The carrier-wave voltage of a frequency fo provided by carrier-wave oscillator S is continuously supplied to back contact 16 with an amplitude sufficient for fully loading the second limiter 10.

The device so far described operates as follows:

In the absence of speech signals, the back contact 16 of relay 9 is closed due to the absence of an envelope voltage. The carrier-wave voltage of 60 lic/S. used as the pilot frequency is supplied to the limiter 10 Via the back contact 16.

As soon as speech frequencies and hence letter sounds are present, speech frequencies pass the filter 13 and bring about an output voltage on low-pass filter 14 Via envelope detector 12. Since the response sensitivity of relay 9 is chosen very low, the relay 9 responds almost immediately upon occurrence of a letter sound, resulting in interruption of back contact 16 and closure of make contact 8. The supply of the pilot voltage to limiter is then interrupted and instead thereof the single sideband speech signal derived from limiter 7 is supplied to limiter 10. As soon as a letter sound has practically died away, the relay 9 is again released.

In the numerical example chosen, the pilot frequency is 60 kc./s. and the frequency-band of the speech signal extends from 60.3 to 63 kc./s. As previously mentioned, the single sideband filter 11 has a pass-band of from 59.9 to 63 kc./s., so that either the limited single sideband signal, or the pilot voltage provided by carrier-wave oscillator 5 occurs at the output of filter 11 as a function of the instantaneous presence or absence of letter sounds. The alternately occurring pilot 'and speech signals have the same amplitude due to the action of limiter- 10.

The signals obtained as described in the foregoing are amplified according to requirements in an amplifier 17 and at the same time transposed by means of a crystal-controlled local oscillator 18 to the desired transmitting frequency and transmitted via an aerial 19.

Since the pilot and speech signals occurring alternately have exactly the same amplitude due to the action of limiter 10, it is possible for the subsequent transmitter stages and more particularly the final stages ofthe transmitters to be so proportioned as to be invariably loaded substantially or completely (class A, B or C-adjustment). In contradistinction with the situation with normal single sideband transmitters, in which the control depends on instantaneous value of the speech signal, the device according to the invention always permits of transmitting with the maximum transmitting power, which is very important with a view to obtaining a high transmission output.

It should be noted that in connection with the described operation of the switching voltage channel for obtaining the switching voltage, it is frequently desirable for the speech signals, -before being supplied to the amplifier 3, to be brought approximately to a given level by utilizing, for example, moderate compression in order to avoid excessive differences in level of the speech signals to be transmitted. For such compression use may be made of level control devices which are known per se for this purpose.

FIG. lb shows a block diagram of a Frenac receiver for use with the Frenac transmitter shown in FIG. la.

The signals received by an aerial 2t) are transposed to an intermediate frequency of 60 kc./s., which for the salte of simplicity has been chosen equal to the intermediate frequency used at the transmitting end, in a superhetrodyne receiver input of the conventional kind preferably comprising a satisfactory AVC-control. The selective pre-amplifier or pre-ampliers and mixing stage or mixing stages required therefor is or are indicated by 21 and the associated local oscillator by 22. The intermediate-frequency and low-frequency portions of the receiver following the input stage are shown in more de tail and comprise an input single-sideband filter 23, the pass-band of which corresponds to that of the single sideband filter 11 of the transmitter and hence is from 59.9 to 63 kc./s. The pilot and single sideband signal which has been filtered out is supplied, after amplification in an intermediate-frequency amplifier 24, to a further single sideband filter 25, the pass-band of which governs the speech frequency-band and, if desired, the pilot frequency and hence is from 60.3 to 63 kc./s. and from 59.9 to 63 kc./s. respectively. The pass-band of this filter has been chosen in conformity with the pass-band of single sideband filter 6 of FIG. 1a. This speech-frequency single sideband is limited by means of a limited 26 for restricting the influence of noise and other interference signals. The limited single sideband speech signal is subsequently supplied via a back contact 27 of a switching relay 28 to a single sideband demodulator 29, which is connected to a local carrier-wave oscillator 30 of 60 kc./s. The low frequency speech signal obtained by demodulation is supplied via a low-pass filter 31 having a cut-olf frequency of about 3 kc./s. and through a low-frequency amplifier 32 having a de-emphasis network to a loudspeaker 33.

The single sideband signal derived from intermediatefrequency amplifier 24, which signal also contains the pilot frequency of 60 kc./s., is supplied via a filter 34 tuned to the pilot frequency to a mixing stage 35, connected to a local oscillator 36. The local oscillator 36 has a frequency of 56 kc./s., so that, together with the frequency of the pilot voltage, a difference frequency of 4 kc./s. is obtained, which may be ltered out with a band-width of about c./s. by means of a simple lter 37. The 4 kc./s. pilot voltage thus filtered out is modulated by the switching voltage. Detection of the pilot voltage is elected by means of amplitude detector 38, and the resultant switching voltage being derived from the output side of a low-pass filter 39 having a cut-off frequency of, for example, 400 c./s., for obtaining suicient suppression of the 4 kc./s. pilot voltage. Similarly as in the transmitter, this switching voltage, if desired after being amplified in a direct-Voltage amplifier, serves to control a switching relay 28. When the pilot voltage is received, which is the case in the absence of a letter sound, switching relay 2S responds and interrupts by means of back contact 27 the speech signal cascade 25 to 33 of the receiver. It is thus ensured that the speech signal cascade passes incoming signals only if letter sounds are present, whereas during speech intervals and even at small intervals between letter sounds the speech signal cascade is interrupted and the noise and other interference signals which then occur cannot reach the loudspeaker 33.

In the receiver shown in FIG. lb, the level of the switching voltage in the switching voltage channel lt-3% is preferably chosen in connection with the response sensitivity of switching relay 28 to be such that the switching relay 28 responds if the instantaneous value of the incoming switching voltage for the weakest signals to be received is about 50% of the maximum instantaneous value. This adjustment affords the advantage that in the absence of the switching voltage, noise voltages then received normally do not cause response of the relay 2S. On the other hand, during reception of the maximum switching voltage, the relay 2S is de-energized only if, as a result of interfering voltages, the incoming switching voltage decreases below 50% of this maximum value.

In the receiver shown in FlG. lb, with suicient preamplification, the switching relay Ztl fulfills the function of a bilateral limiter (Slicer) at a level of 50% of the maximum switching voltage. With the choice of the band-width of the filters used in the switching voltage channel 34-39, it is necessary to make allowance for the fact that due to the action of relay 28 or a similar means for bilateral limitation, the apparent band-Width of th iltering means in the switching voltage channel is decreased. lf the band-width for the amplitude-modulated Signal is lOG c./s. and the cut-off frequency of low-pass iilter 39 is about 50 c.s. for a flank steepness of 6 db per octave, switching voltage frequencies up to about l() c./s. are nevertheless transmitted due to the limiting action of switching relay 28.

The operation of the switching voltage channels in the transmitter and the. receiver of FIG. l will now be explained more fully with reference to the voltage-time diagrams shown in FIGS. 25J-f.

FIG. 2a shows an oscillogram for the spoken syllable tik of the German work Akustik The envelope form of this syllable which consists of three letter sounds pronounced separately is shown in FIG. 2b. As may be seen from this figure, the frequency for the occurrence of separate letter sounds is higher than the so-called syllable frequency. 0n the average, the letter sound frequency is more than twice that of the syllable frequency.

'.In FIG. 2b, dotted line q indicatesr the response level for switching relay 9. As soon as the envelope voltage, which occurs, for example, at the output of the low-pass filter 14 of FIG. 1a, passes the response level q in the upward direction, the switching relay 9 responds with a delay negligible for the signals concerned. The relay 9 remains energized till the envelope signal passes the level q in the downward direction. In this Way, the switching voltage shown in FIG. 2c occurs, as a function of which the

contacts

8 and 16 of switching relay 9 in the Frenac transmitter are opened andclosed.

The described switching voltage varies between two fixed values, for example 0 and l, yand is transferred as a modulation of the pilot signal shown in FIG. 2a' to the Frenac receiver. However, as a result of noise and the like, the incoming voltage shows a curve as shown in FIG. 2e, which greatly differs from the switching voltage of FIG. 2c. Such -a voltage occurs, for example, at the output of low-pass lilter 39 in FIG. lb. As previously described, the response threshold of switching relay 28 is chosen at about 50% of the maximum switching voltage, which is indicated in FIG. 2e by the dotted limit- -ing levels r. The incoming distorted voltage then brings about alternate energization of switching relay 28 in accordance with the curve shown in full line in FG. 2f, so that the elect of interference superposed on the switching voltager is reduced to a minimum. As soon yas a switching voltage having a value l is received, this results in;

the speech signal cascade 25 to 33 being cut off due to interruption of the back contact 27 of switching relay 28.

Reception of speech signals with Frenac equipment of Ithe kind described has been found possible with very bad signal-to-noise ratios `for the single sideband signal, for example about 3 db. In this connection allowance must be made for the fact that, when the single sidebandr signal has a band-width of 3 ken/S., the signal-to-noise ratio for the switching voltage transmitted with a narrow band is considerably more favourable. If the bandwidth of the switching voltage channel is about c./s., the signal-to-noise ratio, for the said numerical value of the single sideband signal, is about l5 db better or, in other words, it is about 18 db. For such a comparatively favourable signal-to-noi-se ratio, the effective value of the noise voltage is small with respect to the mean incoming switching voltage with a resultant possibility of interference for the switching voltage, which is less than .1% (one per thousand), and which means unwanted occurrence of a switching volt-age pulse on the average oncc per l0 seconds. This possibility of interferencel is too low to aect the audibility of speech, so that in the numerical example given, only the inuence of noise upon the limit ed single sideband signal is important. The latter inuence is limited, however, by clipping of the incoming single sideband speech signal.

With sign-al-tonoise ratios lower than 3 db for the single sideband signal, for example 0 db, and a preportionally less favourable signal-to-noise for the switching voltage channel, the possibility of interference in the switching voltage channel considerable increases, since now the effective value of the noise in the. switching voltage channel more nearly approaches the response threshold having a value of about 50% of the maximum switching voltage and hence noise peaks can reach and exceed this threshold more easily. Consequently, with the said signal-to-noise ratio of O db for the single sideband signal, a kind of simmering occurs in the receiver output, which does not prevent speech communication, but which brings about granularity of the incoming speech signal.

For even lower signal-to-noise ratios, the possibility of interference for the switching voltage increases very rapidly, whilst furthermore, due to the action ofthe limiter, instead of suppressing the interfering noise., .the desired weaker speech signal is now suppressed more or less. Both elfects together cause the speech transmission to become rapidly unusable at signal-to-noise ratios lower than 0 db for the single sideband signal, although the correlation of the various frequencies present inl the speech conteracts unusability of the speech communication.

FIGURES 3a and 3b show a Frenac transmitter and a Frenac receiver, respectively, in substantially detail circuits, which equipment permits speech communication up to signal-to-noise ratios for the single sideband signal of about 0 db. The described detail circuits largely correspond to the block diagrams shown in FIGS. la and lb, but these parts of the equipment which are not essential to a proper understanding of the invention are omitted.

In the transmitter device shown in FIG. 3a, the speech signals `amplified and, if desired, compressed are supplied via input terminals 34 and an input transformer 35 to the control-grid of a pentode amplifier 36. The input circuit of transformer 35 includes a seriesfresistor 37, whilst the output circuit includes a seriesfcapacitor 3S and a shunt resister 39. The resistor 37 serves tok benefit high speech frequencies with respect to low speech frequencies (pre-emphasis). The RC-

combination

38, 39 gives an additional damping for frequencies below 300 c./s. This Ril-combination aims lat a bilateral object, firstly, substantial suppression of hum voltages of mainsl frequency, lest they give rise in the envelope detector to the indication signal and, secondly suppression of low 9. speech frequencies, lest speech frequencies reach the narrowibandpass lter for the switching voltage used at the receiving end, so that speech frequencies could wrongly simulate the presence of a pilot signal.

Two channels are connected to the anode circuit of pentode amplifier' 36, viz. a single sideband channel and a switching voltage channel.

The single sideband channel includes a pushpull modulator 49 having an input and an output transformer 4l, 42, respectively. Via centre tappings on these transformers, a carrier-wave of 60 kc./s. `derived from a local oscillator 43 is sup-plied to pushpull modulator 40. A single sideband filter 44 having a pass-band of from 60.3 to 63 kc./s. is connected to the output transformer 42 (not shown in detail). The single sideband filter 44 suppresses tany carrier-wave leak of the push-pull modulator, together with the unwanted lower sideband. The output circuit of single sideband iilter 44 includes a limiter in the form of a diode 46 which is biassed by a battery 45. The resultant limited single sideband signal is supplied via a coupling capacitor 47 Ito the control grid of a triode 48, which triode, together withtriode 49, constitutes a relay circuit which will be described hereinafter.

The anode circuit of pentodearnpliiier 36 furthermore includes a tuned anode circuit Se, which constitutes the input circuit for the'switching voltage channel which will now be described. Circuit 50 has been chosen of low quality to ensure thatja comparatively broad frequencyband of, for example, 0.5 to 3 kc./s. is passed to a coil 51 coupled to this circuit. It isl to be noted that it is not necessary for speech frequencies below 500 c./s. to be supplied to the switching voltage channel, since Vupon occurrence of letter sound frequencies are always present, in the said frequency-band of 0.5 to 3 kc./s., whilst the supply of lower frequencies to the switching voltage channel might lead .to the switching voltage channel being influenced -by interference voltages.

t The speech voltages derived from coupling coil 5l are supplied to an amplitude detector 52 including an output resistor 54 shunted by a capacitor 53. The unipolar envelope obtained by detection is led via a lowpass filter having a series-resistor 5S and a shunt capacitor 56 to the control grid of a pentode amplifier 57, which pentode is connected as a direct-voltage amplilier and amplities the switching voltage to from 50 to l() volts. The anode of pentode 57 is galvanically coupled to the control grid of triode 49, which forms part of' the aforementioned relay circuit comprising triodes 48, 49. The anode of pentode 57 is furthermore connected via

anode resistors

8, 59 to the positive terminal 6i) of an anode voltage source (not shown). A voltage of 60 kc./s. provided by carrier-Wave oscillator 43 is applied via a coupling capacitor 60 to the junction of lthe anode resistors S, 59.

' The relay

circuit comprising triodes

43, 49 is of a type known per se. The common cathode lead of the two triodes includes a cathode resistor 61, shunted by a capacitor 62. The anodes of the -two triodes are directly connected together and connected to the positive anode voltage terminal `60 via an output circuit 63 and an anode resistor 64 with a decoupling capacitor 65.

Either the triode 48, or the triode 49 of the relay circuit is conducting as a function of the polarity of the potential difference between the control gridsv of the two triodes. The right-hand triode 48 acquires a biassing potential via a voltage divider having resistors `66, 67, a smoothing capacitor 68 connected parallel to the lastmentioned resistor and a series-resistor 69. Said biassing potential is, for example, 10G volts.

rIhe control grid of the left-hand triode 49 is connected directly to theanode of pentode57, which anode has, for example, a voltage of ll() volts, provided that the pentode conveys a' low anode current due to the automatic negative grid-bias derived from a cathode resistor 70, this in the absence of an output volta-ge across detector resistor 54.

In the case under consideration pentode 57 causes the potential of the control grid of the left-hand triode 49 to be higher than the potential of the control grid of the right-hand triode 48, so that triode 49 is conducting and triode 48 is cut olf. Via coupling capacitor 60 and the junction of the

anode resistors

58, 59 the control grid of the left-hand triode has applied to it a pilot voltage of suitable amplitude, which reaches via triode 49 the output circuit 63 of relay circuit 48, 49. This condition occurs in the absence of a voltage across the output circuit of amplitude detector 52-54, in other words so long as letter sounds are not supplied to pentode amplifier 36.

As soon as a letter sound occurs, amplitude detector 52-54 detects the envelope thereof, this envelope after being smoothed by the low-pass filter 55, 56 (cut-off frequency, for example, from to 200 c./s.) controlling the control grid of pentode amplifier 57; This leads yto the occurrence of a considerable anode current of pentode 57 and a corresponding decrease in the anode voltage thereof, so that the potential of thel control grid of triode 49 Idecreases below the potential of the control grid of triode 48. Consequently, .triodef49 is cut ,oil` and triode 48 becomes conducting. "Via coupling capacitor 47, connected to the single lsideband modulatorY 4l),- the speech `signals occurring are supplied to the 'control'- grid of triode 48 and hence to output circuit 63 f the relay circuit.Y v

As soon as an interruption of, for example, from l0* to 2.0 msecs. occurs between the letter sounds supplied to the input terminals 34 and hence, -for example, during the interval between sequential letter sounds, pronounced separately, of the speech signals supplied, the output voltage ofthe envelope-detector rapidly decreases to such an extent that the relay circuit 4S, 49 returns to its initial condition, in which triode 49 is conducting and triode 48 is cut oft. As before, only the pilot voltage then occurs across the output circuit` 63 -of `the relay circuit. It should be noted that the time-constant of the cathode circuit of the relay circuit must be chosen sufciently small, for example 0.1 msec. to obtain sul-liciently rapid action.

As previously described with reference to FIG. la,

the switching voltage channel including envelope detector 52-54 must be so proportioned that the relay circuit responds already upon occurrence of speech signals which are small with respect to the maximum value thereof and more particularly less than 5%, for example 2%. Such very early response of the switchingl voltage channel prevents the partial loss of fractions of letter sounds, for example in the starting range of letter sounds, which are necessary for proper audibility of the speech diode 46 may be designed in the simple manner as shownfv since only moderate limitation is required therein. The output signals which occur across limiter 70 are supplied via a coupling capacitor 71 and a single sideband lter 72 having output terminals 73, to the further transmitting equipment (not shown). As a matter of fact, the single'sideband filter 72 must have a pass-band suficient toY pass on the one hand, the pilot signal of 60 kc./s. which is modulated by the switching voltageand, on the other hand, the single sideband speech' signal con- 1 1 taining ,the maximum speech frequencies of about 3 kc./s., the pass-band may be, for example, vfrom 59.8 to 63 kc./s.

It willbe evident that in FIG. 3a the relay circuit comprising triodes 48, 49 is the equivalent ot' the switching relay 9 with

contacts

8 and 16 of FlG. la. Although it is possible, as shown in FIG. la, to utilize a quick-switching relay, it is more advantageous, as shown in FlG. 3a, to utilize an electronic relay circuit which has substantially no inertia.

FIG. 3b shows. a portion of a Frenac receiver in detail, thatis to sayy an intermediate-frequency portion and a demodulator portion, beginning with input terminals '75 connected to a single sideband lter '74. If again the frequency of the suppressed carrier wave is 60 kc./s., said single sideband lter has a pass-band, from 59.9 to 63 kc./s. It thus passes not only the single sideband speech signals, but alsov the pilot voltage of 60 kc./s. which is modulated by the switching voltage, but the bandwidth for the modulated pilot voltage at the receiving side has been chosen to bek smaller than that at the transmitting end. The output ofthe single sideband filter 74 is connected to the control grid of an amplifying pentode 76 having two anode circuits. connected in series, 77 and 78 respectively.

Anode circuit 78 is tuned to central frequency of the speech sideband andk is intended to transmit this speech sideband to the single sideband demodulator. For this purpose, the (carrier wave and) single sideband speech signals, which occur across circuit 78 are supplied via a limiter circuit having a diode 79 and a biassing resistor 80 andfurthermore a couplingcapacitor 81 to the control grid of a triode S2 which, together with a diode 83, constitutes a relay circuit of the same type as the relay circuit having triodes 4S and 49 which was described with reference to FG. 3a. The biassing resistor 80 belonging to the limiter 79, 80 is included in the connecting lead from the terminal 84 for the positive anode voltage tothe end; of circuit '7S- which is remote from the anode of pentode 76. A smoothing capacitor 85k is connected to resistor I80.

The objectJ of the` anode circuit 77 is to lter out the modulated pilot voltage. and for this purpose it is tuned to the pilot frequency of 60 ke./s. Via a coupling coil 86 the pilotvoltage ltered out is supplied in series with a voltage. of a frequency of 56 kc./s. `provided by an auxiliary oscillator 87 to the central grid of a pentode 88 which actsv ask a mixer- The anode circuit of pentode 88 includes a circuit 89 which, is tuned to the different frequency of the` voltages supplied and hence to 4 kc./s., the band-width of this circuit being about 100 c./s. The pilot voltage thus sharply filtered outis led via a coupling coil 90 to an amplitude detector havingv a diode 91 andv an output resistor 93 bridged by a shunt capacitor 92. The switching voltage obtained by this detection, after being smoothed by a smoothing filter having a seriesresistor 94 and a shunt capacitorl 95, is applied to the control grid of triode 83.

The triodes 82 and 83' have a common cathodey impedance comprising a cathode resistor 96 and a parallel capacitor 97 and an output transformer 9-8 included in the common anode. lead. Ina similar manner as in the relay circuit shown in FIG. 3a, the control grid of triode 82 acquires by means of a voltage divider comprising resistors 99, 163, 101 and a decoupling capacitor 162 a biassing potential ofk about 100 volts via series-resistors 163, 104i. The control-grid of the left-hand triode 83 acquires a biassing potential derived from potentiometer resistor 101, which biassing potential isV thus less than that for the control-grid ofthe right-hand triodeV 82. Consequently,` in the absence of a voltage across detector resistor 93, triode 821is conducting and triode 83 is cut olf. The singlesideband signals supplied via coupling capacitor 81 to the control grid of triode 82 are then transmittedv to theoutput transformer 9%l and demodulated in a push-pull demodulator 10S by means of a voltage havinga frequency of 60 kc./s. derived from a local carrierwave oscillator 106. Via an output transformer 107, an output resistor 103 and a smoothing capacitor 109 which serves inter alia for de-emphasis, the speech signals obtained by demodulation, after being smoothed further by means of a iilter 11i), are supplied via output terminals 111 to a low-frequency amplifier (not shown).

When a pilot voltage is received, this results in the occurrence of an output voltage on amplitude detector 91-93, so that the potential of the control grid of triode 83 becomes higher than that of the control grid of triode 82. The relay circuit then changes over, which prevents the single sideband signal applied to the control grid of triode 82 from being transmitted. The switching voltage applied to the control grid of triode r83 cannot reach via demodulator the output terminals 111, since this switching voltage contains only comparatively low frequencies. As soon as a. pilot voltage is no longer received, the potential of the control grid of trigger 83 decreases to its initial value and the relay circuit returns to its initial position, single sideband speechvsignals again being transmitted to output terminals 111.

With the described circuit, itis thusv ensured that with alternatey reception. of speech signals and the pilot signal the speech signal cascade 78--82, 98--105, 1417-111 is each time released only upon reception of single sideband speech signals, but is cut off when a pilot signal is received, for example in speech intervals, intervals between words and also with very transient intervals between letter sounds.

Optimum operation of the relay circuit comprising the

triodes

82, 83 is obtained if it responds to about 50% of the maximum value of the incoming switching voltage. 1n order to make this favourable` adjustment more or less independent of fading phenomena, it is desirable for the receiver portionsv preceding` the described circuit to be provided with a rigidly operating AVC-circuit, in order to bring the maximum switching voltage to a predetermined level.

Due to the small band-Width of the tuned circuit 89 which serves to select. thel pilot signal, of the circuit of amplitude detector 91-93, and of the

subsequent lowpass lter

94, 95,r the switching voltage supplied to the control grid of triode 83 has noy rectangular shape, its angles being rounded and its flanks being limited in` steepness. As a rule, it is advantageous to design the relay circuit 82, S3 in a manner such, inter alia by means of suflicient pre-amplification of the switching voltage, that it responds veryv rapidly and unwanted retardations in making the speech signal cascade operative are avoided. With respect to the detected switching voltage,v the relay circuit then acts as a bilateral limiter and the speech signal cascade is switched on and olf very abruptly. It has been found thatthis abrupt switching on and olf detracts from the audibility and recognizability of voices. This disadvantage may be obviated in part by means of suitable rounding of the switching voltage and for this purpose a retarding capacitor 112 is included between the junction of the series-resistors y103 and 104 in the control-grid circuit'of triode S2 and the common cathode lead. The time-constant of the retarding network resulting from the use of said capacitor 112 in the control-grid circuit of triode 32 may be chosen equal to about 2 msecs., in which event the cut-off frequency is about equal to the lowest speech frequency transmitted. This results in the speech signal cascade no longer being switched on and olf in accordance with a voltage of rectangular shape,

but in accordance with a voltage which hasv a form as indicated in dotted line in FIG. 2f. The envelope of letter sounds passed via the speech signal cascade then acquires a more natural forml uponv swelling and dying away (cf. FIG. 2a).

It will be` evident that, several variants of the described embodiments are possible within the scope of the inven- 13 tion. Thus, for example, the relay circuit used in FIGS. 3a and 3b, which also fulfills the function of a bilateral limiter, may be designed as a bilateral limiter with suitabily biassed diodes, followed by static relays for switching the various channels Iinto and out of circuit.

For amplifying the rectangular switching voltage, use may be made of non-linear ampliiiers. Thus, for example, a trigger circuit, more particularly a bistable multivibrator circuit for amplifying and bilateral limitation of the incoming switching voltage may be included in the receiver of FIG. 3b between the smoothing

lter

94, 95 for the switching voltage and the left-hand triode 83 of the relay circuit. Furthermore, the described electronic relay circuit itself may be changed to a relaxation circuit having two stable positions of equilibrium. The limiting level in this case may be chosen to be fixed as described above or, if desired variable in proportion to the receiving strength of, for example, the incoming pilot voltage in order to minimize the influence of noise upon the regenerated switching voltage.

In the embodiments of the receiver as described, the interrupting switch is included in the high-frequency porton of the speech signal cascade; the interrupting switch may alternatively be included in the low-frequency portion of the speech signal cascade, but in this case particular steps must be taken for suppression of audible switching frequencies.

For the alternate transmission of a pilot signal and speech signals, it is not necessary to utilize relay circuits having a change-over switch as are present in the transmitters shown in FIGS. la and 3a. For example, the required action may also be obtained by means of a circuit as shown in FIG. la, if a pilot signal of suitable amplitude is introduced between the lirniters 7 and 10 which in this case are through-connected, viz. in such manner that at the input of limiter 10 the pilot signal is .about 35 db weaker than the speech signal, but sufticiently strong for complete driving of the limiter 10. In this case, without the use of a switching voltage channel as indicated at 12-15 in FIG. la the pilot signal completely loads the limiter, and the following transmitting equipment in the absence of speech signals, whereas in the presenceof speech signals the pilot signal is suppressed almost completely due to the action of the limiter 10, viz. is in the output circuit of limiter 10 more than 40 db weaker than the single sideband speech signals. With the chosen ratio of pilot signal and single sideband speech signal at the input of limiter 10, the pilot signal is abruptly suppressed as soon as a letter sound is received and again occurs abruptly during dying away of the letter sound.

Furthermore, if frequency-diversity transmission for the switching voltage is desired, not only a pilot voltage of 60 kc./s., but also a pilot voltage of, for example, 63.5 orV 64 kc./s. may be applied to the back contact 16 of FIG. 1b or to the control grid of relay circuit triode 49 of FIG. 3a. As a matter of fact, several of the single sideband lters used (11 in FIG. la, 72 in FIG. 3a) must in this case have a pass-band which passes both pilot frequencies.

In the embodiments described, the pilot signal modulated bythe 1 switching voltage and the limited single sideband signal are transmitted alternately. It is possible to transmit the pilot signal and the limited single sideband signal simultaneously, but in this case a strongly pulsating and varying load on the transmitter results and this is normally not desirable for transmitters of the kind concerned.

For the sake of completeness, it is pointed out that, when using frequency-diversity and/or when using spatial diversity at the receiving end, an advantageous signalto-noise ratio for the incoming switching voltage is liable to be generally affected if no particular steps are taken, since the noise voltages of the various switching voltage channels must be summated in determining the signal-tonoise ratio. In view thereof, when use is made of diversity, it is necessary to aim at an equipment in which each time only the switching voltage channel with the optimum receiving conditions is active and the others are cut olf and hence cannot add to the noise in the output of the switching voltage channel. In this case only can full profit be taken from the possibility of speech communication with exceptionally low signal-to-noise ratio as afforded by the invention.

What is claimed is:

1. A single-sideband system comprising a transmitter having means for producing a single-sideband signal controlled by speech signals, a limiter connected to limit the amplitude of said single-sideband signal, means for producing a pilot signal switching Voltage which varies discontinuously between fixed values in accordance with the instantaneous presence and absence of said speech signals, and means for transmitting said limited singlesideband signal and said pilot signal switching voltage, and a receiver having means to receive said transmitted signals, a single-sideband demodulator, a limiter, means connected to apply the received single-sideband signal to said demodulator through the last-named limiter, an output means connected to said demodulator, and a switching means controlled by the received switching voltage and connected to permit the received signal to reach said output means only during the presence of said speech signals.

2. A single-sideband transmitter comprising means for producing a single-sideband signal controlled by speech signals, a limiter connected to limit the amplitude of said single-sideband signal to substantially constant amplitude, means for producing a switching voltage which varies discontinuously between iixed values in accordance with the instantaneous presence and absence of said speech signals, and means for transmitting said limited singlesideband signal and switching voltage.

3. A transmitter as claimed in claim 2, including means for transmitting alternately said limited single-sideband signal and said switching voltage.

4. A single-sideband transmitter comprising means for producing a single-sideband signal controlled by speech signals, a limiter connected to limit the amplitude of said single-sideband signal, and means for producing a switching voltage which varies discontinuously between fixed values in accordance with the instantaneous lpresence and absence of said speech signals, said last-named means comprising an envelope detector connected to produce an envelope of said speech signals, a smoothing illter connected to the output of said envelope detector, and means connected to the output of said smoothing iilter to derive a switching voltage from said envelope.

5. A transmitter as claimed in claim 4, in which said means to derive a switching voltage comprises a relay connected to be actuated by said envelope.

6. A transmitter as claimed in claim 5, in which said relay is an electronic switching circuit.

7. A transmitter as claimed in claim 4, including a pilot signal generator, and switching means responsive to said switching voltage, said switching means being connected to cause alternate transmission of said single-sideband signal and said pilot signal.

8. A single-sideband receiver for receiving a singlesideband speech signal and a switching voltage which varies discontinuously between fixed values in accordance with the instantaneous presence and absence of the speech signals, comprising a single-sideband demodulator, a limiter, means connected to apply said single-sideband speech signal to said demodulator through said limiter, an output means .connected to said demodulator, and a switching means controlled by said switching voltage and connected to permit the received signal to reach said output'means only during the presence of speech signals.

9. A single-sideband receiver for receiving .a singlesideband speech signal and a switching voltage characterized by a pilot: signal which is present only during the absence of speech signals, comprising a single-sideband demodulator, a limiter, means connected to apply said single-sideband speech signal to-said demodulator through said limiter, an output means connected to said demodulator, and a switching voltage channel comprising successively a pilotffrequency selector connected to receive said pilot signal, a switching voltage detector, and a relay circuit connected to permit the speech signalY to reach said output means only during the absence of said pilot signal. t

10. A receiver as claimed in claim 9, in which said switching voltage channel includes a lter having an upper cut-off frequency which is lower than the lowest speech frequency appearing; at the output of said de- 15 2,359,338.

modulator.

16 11. A receiver. as claimed in claim 9, in which said relay circuit is an electronic switch.

References Cited inthe ile of this patent UNlTED STATES PATENTS 1,948,973 Steinberg Feb. 27, 1934 2,210,957 Skillman Aug; 1-3, 1940 2,215,482 Skillman Sept. 24 1940 2,231,958 Skillrnan Feb. 1.8, 1941V 2,237,899 Bjornson ApuV 8 1941 2,543,807 Saling Mar. 6, 1951 2,632,800 Schlesinger Mar. 24, 1953 2,810,787' DiToro et al. Oct. 22', 1957 Paluka Nov. 4, 1958