DK160171B

DK160171B - CLUTCH TO RECOVER THE BATCH SIGNAL FROM AN NRZ CODED, BINARY MESSAGE

Info

Publication number: DK160171B
Application number: DK217383A
Authority: DK
Inventors: Claude Pivon; Georges Depriester
Original assignee: Sintra Alcatel Sa
Priority date: 1982-05-17
Filing date: 1983-05-16
Publication date: 1991-02-04
Also published as: DK217383D0; DK160171C; EP0095083B1; IE54411B1; DE3369620D1; FR2527031A1; DK217383A; FR2527031B1; EP0095083A1; GR78569B; IE831135L

Description

Opfindelsen har relation til numeriske transmissioner og den beskæftiger sig med genvinding af det taktsignal, der definerer de tidsintervaller mellem hvilke samtlige overgange i en NRZ-kodet binær meddelelse be-5 finder sig.The invention relates to numerical transmissions and it deals with the recovery of the clock signal which defines the time intervals between which all transitions in an NRZ-encoded binary message are located.

Der kendes mange koblinger til genvinding af taktsignalet fra en NRZ-kodet binær meddelelse. Disse koblinger indbefatter sædvanligvis en lokal taktgiver, der arbejder på frekvensen af taktsignalet i den for-10 ventede NRZ-kodede binære meddelelse, og som omfatter en faselåsesløjfe, der søger at bringe overgangene i den binære meddelelse til at falde sammen med periodebegyndelserne. Sløjfen udgøres af en kreds til regulering af fasen af den lokale taktgenerator, hvilken kreds sty-15 res af en komparator, der detekterer positionen af overgangene i den NRZ-kodede binære meddelelse i forhold til det lokale taktsignals periodebegyndelser. Der er indskudt mere eller mindre sofistikerede filtre mellem kom-paratoren og fasereguleringskredsen med henblik på op-20 nåelse af en meget kort indfangningstid og en stor ufølsomhed over for de forstyrrende signaler.Many couplings are known for recovering the clock signal from an NRZ-encoded binary message. These couplings usually include a local clock sensor operating on the frequency of the clock signal in the expected NRZ-encoded binary message, and which includes a phase lock loop which seeks to coincide with the transitions in the binary message. The loop is constituted by a circuit for regulating the phase of the local clock generator, which circuit is controlled by a comparator which detects the position of the transitions in the NRZ-encoded binary message relative to the period start of the local clock signal. More or less sophisticated filters are interposed between the comparator and the phase control circuit in order to achieve a very short capture time and a high sensitivity to the interfering signals.

Et udførelseseksempel på sådanne koblinger fin-r des beskrevet i US-A-3.851.101.An embodiment of such couplings is described in US-A-3,851,101.

Genvindingen af taktsignalet fra en binær medde-25 lelse er en væsentlig egenskab navnlig i det tilfælde, hvor de NRZ-kodede binære meddelelser forekommer i form af asynkrone tog, hvor enhver forsinkelse i genvindingen af taktsignalet kan føre til tab af data eller til at der må anvendes et bufferlager med meget stor kapacitet. 30 Genvindingen af taktsignalet skades dog af det forhold, at der til faseregulering af den lokale taktgiver, findes en ustabil ligevægtsstill±ng,hvor det lokale taktsignal er i modfase med taktsignalet i den modtagne meddelelse, og hvor reguleringen kan opretholdes en hvis 35 tid ved en rækkefølge af korrektioner med skiftevis fasefremrykning og faseforsinkelse.The recovery of the clock signal from a binary message is an essential feature, especially in the case where the NRZ-encoded binary messages occur in the form of asynchronous trains, where any delay in the recovery of the clock signal may lead to data loss or may be used a buffer capacity with very large capacity. 30 However, the recovery of the rate signal is damaged by the fact that for phase control of the local rate sensor there is an unstable equilibrium position, where the local rate signal is in phase with the rate signal in the received message and where the regulation can be maintained if 35 times at a sequence of corrections with alternate phase advancement and phase delay.

DK 160171 BDK 160171 B

22

Hos de fleste af de kendte koblinger til genvinding af taktsignal undgår man den nævnte ustabile ligevægtsstilling i fasereguleringen ved at kontrollere, at der i den modtagne meddelelse ikke forekommer overgange 5 i nærheden af midten af det lokale taktsignals perioder, og ved i modsat fald at foretage en pludselig fasekor-rigering, hvorved fasereguleringen bringes ud af den labile ligevægtsstilling.In most of the known rate signal couplings, the said unstable equilibrium position in the phase control is avoided by checking that in the received message, transitions 5 do not occur near the center of the local rate signal periods, and if not, a sudden phase correction whereby the phase control is brought out of the labile equilibrium position.

I praksis er det nødvendigt at tage hensyn til 10 støjen,og derfor må man vente til at der i den modtagne meddelelse forekommer flere overgange, der i hovedsagen står i modfase med det lokale taktsignal,inden man med sikkerhed kan træffe afgørelse om at faselåsesløjfen skal bringes ud af låsning. Som følge heraf er der en 15 svag men ikke ubetydelig sandsynlighed for at der foregår et langvarigt tidsrum til genvinding af taktsignalet, således at en del af den modtagne meddelelse går tabt.In practice, it is necessary to take into account the noise, and therefore one must wait for the received message to have several transitions, which in the main proceedings are at odds with the local clock signal, before a decision can be made that the phase lock loop should be decided. brought out of lock. As a result, there is a slight but not insignificant probability that a long time period is needed for the recovery of the clock signal, so that part of the received message is lost.

For at undgå dette er der mange af de kendte kob-20 linger, hvor der systematisk foretages afprøvning af stabiliteten i den første ligevægtsstilling, der opnås, ved på det lokale taktsignal at påtrykke en pludselig faseændring. Dette sker dog på bekostning af gennemsnitsvarigheden af taktsignalgenvindingen med praktisk 25 taget en fordobling af denne varighed.To avoid this, there are many of the known couplings in which the stability of the first equilibrium position obtained is systematically tested by applying a sudden phase change to the local clock signal. However, this is done at the expense of the average duration of the rate signal recovery with practically a doubling of this duration.

Opfindelsen tager sigte på at forkorte tidsrummet for genvinding af taktsignalet fra en NRZ-kodet meddelelse ved i starten af en genvindingsprocedure, at undgå muligheden for opnåelse af en labil ligevægt for 30 fasereguleringen af det lokale taktsignal, hvilket opnås ved en forudindstilling af det lokale taktsignals fase på basis af en mere eller mindre grov vurdering af fasen af den modtagne meddelelses taktsignal, medens denne meddelelse føres igennem en forsinkelseslinie.The invention aims to shorten the period of recovery of the clock signal from an NRZ-encoded message by avoiding, at the beginning of a recovery procedure, the possibility of achieving a labile equilibrium for the phase control of the local clock signal, which is achieved by a presetting of the local clock signal phase on the basis of a more or less coarse evaluation of the phase of the received message's clock signal, while passing this message through a delay line.

35 Med henblik herpå er en kobling til genvinding35 For this purpose, a coupling is for recovery

DK 160171BDK 160171B

3 af taktsignalet fra en NRZ-kodet binær meddelelse, og af den art, der omfatter rytmegivende midler til frembringelse af et rytmesignal, der har en frekvens lig med et multiplum p af frekvensen af den binære meddelelses 5 taktsignal, og som definerer successive tidsvinduer af ens varighed, tællemidler, der opererer på de over flere taktperioder modtagne signalovergange, og maksimumsdetekteringsmidler, der opererer på de opnåede tælleværdier ifølge opfindelsen ejendommelig ved, at tællemid-10 lerne er indrettede til at gruppere tidsvinduerne i klasser afhængigt af deres modulo p-rang i forhold til et arbitrært udgangstidspunkt, og til indenfor hvert tidsvindue at fastlægge det antal overgange, der er faldet sammen med tidligere tidsvinduer af samme klasse modulo 15 p som det igangværende tidsvindue, og at maksimumsdetekteringsmidlerne er indrettede til at afgive et taktinitialiseringssignal i de tidsvinduer, der hører til den modulo p-klasse, der har den maksimale tælleværdi, hvorhos koblingen desuden har en lokal taktgenerator, der 20 har den modtagne meddelelses taktfrekvens, og som synkroniseres ved begyndelsen af genvindingsproceduren ved hjælp af maksimumsdetekteringsmidlerne.3 of the clock signal from an NRZ-encoded binary message, and of the nature comprising rhythmic means for generating a rhythm signal having a frequency equal to a multiple p of the frequency of the binary message 5 clock signal and defining successive time windows of one duration, counting means operating on the signal transitions received over multiple clock periods, and maximum detecting means operating on the obtained count values according to the invention peculiar in that the counting means are arranged to group the time windows into classes depending on their modular p relative to an arbitrary exit time, and to determine within each time window the number of transitions that coincided with previous time windows of the same class modulo 15p as the current time window and that the maximum detection means are arranged to output a clock initialization signal in the time windows which belongs to the modulo p class that has the maximum count value, the coupling further having a local clock generator having the clock frequency of the received message, which is synchronized at the beginning of the recovery procedure by the maximum detection means.

Opfindelsen angår også en kobling til genvinding af taktsignalet fra en NRZ-kodet binær meddelelse, og 25 af den art, der omfatter rytmegivende midler til frembringelse af et rytmesignal, der har en frekvens lig med et multiplum p af frekvensen af den binære meddelelses taktsignal, og som definerer successive tidsvinduer af ens varighed, optællingsmidler, der opererer på de over 30 flere taktperioder modtagne signalovergange, og minimumsdetekteringsmidler, der opererer på de fremkomne fordelingsforskelle, hvilken kobling ifølge opfindelsen er ejendommelig ved, at optællingsmidlerne er indrettede til at gruppere tidsvinduerne i klasser afhængigt af 35 deres modulo p-rang i forhold til et arbitrært udgangstidspunkt og til indenfor hvert tidsvindue at fastlæggeThe invention also relates to a coupling for recovering the rate signal from an NRZ-encoded binary message, and of the kind comprising rhythm-generating means for generating a rhythm signal having a frequency equal to a multiple p of the frequency of the binary message rate signal. and defining successive time windows of equal duration, counting means operating on the signal transitions received over 30 multiple clock periods, and minimum detecting means operating on the resulting distribution differences, which coupling according to the invention is characterized in that the counting means are arranged to group the time windows into classes. depending on their modulo p-rank relative to an arbitrary starting time and to determine within each time window

DK 160171 BDK 160171 B

4 fordelingsforskellen mellem de overgange, der ligger forud, og de overgange, der ligger bagud i forhold til tidsvinduerne af samme klasse som det igangværende vindue, at minimumsdetekteringsmidlerne er indrettede til 5 at afgive et taktinitialiseringssignal i de tidsvinduer, der hører til den klasse modulo p, der udviser den laveste fordelingsfordkel, hvorhos koblingen desuden har verifikationsmidler til spærring af minimumsdetekteringsmidlerne, når fordelingsforskellen for en klasse med 10 modulo p-rang maksimalt er \ p mindre henholdsvis f p større end rangen af den klasse tidsvinduer for hvilke et detekteret minimum i fordelingsforskellen ikke tillader afgørelse til forskel for de forsinkede eller de fremskudte overgange, samt en lokal taktgenerator, der 15 har den modtagne meddelelses taktfrekvens, og som synkroniseres ved begyndelsen af synkroniseringsproceduren ved hjælp af minimumsdetekteringsmidlerne.4 shows the distribution difference between the preceding transitions and the transitions which are lagging relative to the time windows of the same class as the current window, that the minimum detection means are arranged to give a clock initialization signal in the time windows belonging to the class modulo p. exhibiting the lowest distribution advantage, the coupling additionally having verification means for blocking the minimum detection means, when the distribution difference for a class with 10 modulo p-ranges is maximum \ p less respectively fp greater than the rank of the class time windows for which a detected minimum in the distribution difference is not allows determination of difference between the delayed or delayed transitions, as well as a local clock generator having the clock frequency of the received message and synchronized at the beginning of the synchronization procedure by the minimum detection means.

Vurderingen af modulo p-rang af den klasse tidsvinduer,med hvilken begyndelserne af perioderne af den 20 modtagne meddelelses taktsignal falder saimen,giver mulighed for i starten at indstille det lokale taktsignal med en faseforskel, der højst er på 2π/ρ, hvor tallet p vælges således, at denne forskel er tilstrækkelig lille til at undertrykke muligheden for at fasereguleringen nærmer 25 sig den labile ligevægtsstilling, og fortrinsvis for med god sandsynlighed at sikre en korrekt behandling af dataene i den modtagne meddelelse.The estimation of modulo p-rank by the class of time windows with which the beginning of the periods of the 20 received message rate signal decreases gives the possibility of initially setting the local rate signal with a phase difference not exceeding 2π / ρ where the number p is selected such that this difference is sufficiently small to suppress the possibility of phase control approaching the labile equilibrium position, and preferably to ensure, with a good probability, the correct processing of the data in the received message.

Da tælle- eller optællingsmidlerne arbejder med fast tidsrum, kan den forsinkelse, der indføres ved den 30 forudgående vurdering af fasen af taktsignalet i den modtagne meddelelse nemt opvejes ved at lade meddelelsen passere gennem en forsinkelseslinie, der indfører en forsinkelse,som er lig med det faste tidsrum.Since the counting or counting means work over a fixed period of time, the delay introduced by the prior assessment of the phase of the clock signal in the received message can be easily offset by passing the message through a delay line introducing a delay equal to that. fixed periods of time.

Optællingsmidlerne kan hensigtsmæssigt bestå af 35 tællemidler og kalkuleringsmidler, der for tidsvinduerne af samme modulo p-rang som det igangværende tidsvindueThe counting means may conveniently consist of 35 counting means and calculating means which for the time windows of the same modulo p-rank as the current time window

DK 160171 BDK 160171 B

5 angiver fordelingsforskellen ved rekursiv beregning på basis af den fordelingsforskel, der blev kalkuleret under det forudgående tidsvindue og de tidligere indsamlede, nyeste værdier for optællingerne vedrørende for-5 skellige klasser af tidsvinduer.5 indicates the distribution difference by recursive calculation on the basis of the distribution difference calculated during the preceding time window and the previously collected latest values for the counts for different classes of time windows.

Opfindelsen forklares nærmere i det følgende under henvisning til den skematiske tegning, hvor fig. 1 viser et blokdiagram over en kobling til genvinding af fasen af taktsignalet i asynkrone, NRZ-10 kodede binære meddelelser med taktsignaler på samme frekvens, og fig. 2 og 3 to udføreIsesformer for den i fig. 1 viste kobling til genvinding af taktsignalet i overensstemmelse med opfindelsen.The invention is explained in more detail below with reference to the schematic drawing, in which fig. 1 is a block diagram of a coupling for recovering the phase of the clock signal in asynchronous NRZ-10 encoded binary messages with clock signals at the same frequency; and FIG. 2 and 3 are two embodiments of the embodiment shown in FIG. 1 to recover the clock signal according to the invention.

15 Den i fig. 1 viste kobling har en indgang 1 for de asynkrone binære meddelelser og et antal udgange 2, 3 og 4. Udgangen 2 afgiver de til indgangen 1 ankommende meddelelser med en konstant forsinkelse, der indbefatter tiden til genvinding af faserne af disse 20 meddelelsers taktsignal, medens den anden udgang 3 afgiver et binært signal D for modulationsdetektering, hvilket signal tilkendegiver om der er en binær meddelelse på udgangen 2 eller ej, hvorhos den tredje udgang 4 afgiver det taktsignal, der er genvundet fra 25 den binære meddelelse, der afgives over udgangen 2.15 The embodiment of FIG. 1, an input 1 for the asynchronous binary messages and a plurality of outputs 2, 3 and 4. The output 2 outputs the messages arriving at the input 1 with a constant delay including the time for recovering the phases of the clock signal of these 20 messages, while the second output 3 outputs a binary signal D for modulation detection, indicating whether there is a binary message on output 2 or not, whereas the third output 4 outputs the rate signal recovered from the binary message output over output 2 .

Koblingen indbefatter en lokal taktgenerator 5 med frekvensen F af taktsignalerne i de modtagne meddelelser, hvilket lokale taktsignal er faselåst til overgangene i den binære meddelelse, der afgives over udgangen 30 2. Koblingen omfatter desuden en forsinkelseslinie 6, der er indskudt mellem indgangen 1 og den lokale taktgenerator 5 og opvejer fasegenvindingstiden, samt en kreds 7 til hurtig fasegenvinding, hvilken kreds tjener til modulationsdetektering, til startsynkronisering 35 af den lokale taktgenerator 5 og indirekte til styring af denne generators regulering.The coupling includes a local clock generator 5 having the frequency F of the clock signals in the received messages, which local clock signal is phase locked to the transitions in the binary message delivered over the output 30 2. The coupling further comprises a delay line 6 interposed between the input 1 and the local clock generator 5 and outweighs the phase recovery time, and a circuit 7 for rapid phase recovery, which serves for modulation detection, for starting synchronization 35 of the local clock generator 5 and indirectly for controlling this generator's control.

DK 160171BDK 160171B

66

Den lokale taktgenerator 5 omfatter en fase-låsesløjfe af første orden med to integratorfiltre 54, 55 som indkobles gennem en omskifter 56. Filteret 54 har en tidskonstant, der er tilstrækkelig lille til 5 at fasesynkroniseringen kan opnås i løbet af ca. 10 bit af en meddelelse, men tilstrækkelig lang til at sikre stabiliteten af faselåseprocessen uden for præstationerne af kredsen 7 til hurtig fasegenvinding. Det andet filter 55 har en stor tidskonstant for at opret-10 holde den genvundne fase selv om der forekommer udfald i forbindelsen, men tidskonstanten er dog kompatibel med en glidning på i det mindste det dobbelte af den forventede glidning mellem sendeoscillatoren og modtageoscillatoren. Filteret 54 med lille tidskonstant er i drift, 15 når der ikke detekteres nogen modulation og når der de-tekteres modulation i et tidsrum, der svarer til de første bit af en meddelelse, dvs. det tidsrum, der er nødvendig til opnåelse af en faseforskel lig med nul.The local clock generator 5 comprises a first-order phase-lock loop with two integrator filters 54, 55 which are engaged through a switch 56. The filter 54 has a time constant sufficiently small for the phase synchronization to be achieved in approx. 10 bits of a message, but long enough to ensure the stability of the phase locking process outside the performance of the circuit 7 for rapid phase recovery. The second filter 55 has a large time constant to maintain the recovered phase even though outages occur, but the time constant is compatible with a slip of at least twice the expected slip between the transmitting oscillator and the receiving oscillator. The filter with low time constant 54 is in operation, 15 when no modulation is detected and when modulation is detected for a period corresponding to the first bits of a message, ie. the time needed to obtain a phase difference equal to zero.

Dette filter erstattes derefter af det andet filter 55.This filter is then replaced by the second filter 55.

20 På konventionel måde omfatter den lokale takt- generator 5 en oscillator 50, hvis frekvens F' er lig med et helt multiplum r af frekvensen F, hvilken oscillator er koblet til en fasemodulator, der udgøres af en r-deler 51 med en resætningsindgang og 25 med en styreindgang, som gør det muligt at ændre delingsforholdet med ί 1. Taktsignalgenvindingskredsen 7 bruger resætningsindgangen til den oprindelige synkronisering af den lokale taktgenerator. Faselåsesløj-fen bruger styreindgangen til ændring af delingsfor-30 holdet r med ± 1 til efter behov at fremrykke eller forsinke fasen af den lokale taktgenerator 5.Conventionally, the local clock generator 5 comprises an oscillator 50, whose frequency F 'is equal to an integer r of frequency F, which oscillator is coupled to a phase modulator constituted by an r-divider 51 having a reset input and 25 with a control input which allows the partition relationship to be changed by ί 1. The clock signal recovery circuit 7 uses the reset input for the original synchronization of the local clock generator. The phase lock loop uses the control input to change the partition ratio r by ± 1 to advance or delay the phase of the local clock generator 5 as needed.

Faselåsesløjfen i den lokale taktgenerator 5 omfatter en overgangsdetektor 53, hvis indgang får tilført de binære meddelelser fra forsinkelseslinien 35 6, og hvis udgang afgiver impulser af kort varighed, der falder sammen med overgangene i disse meddelelser.The phase lock loop of the local clock generator 5 comprises a transition detector 53 whose input is fed to the binary messages from the delay line 35 6 and whose output emits short duration pulses coinciding with the transitions of these messages.

DK 160171BDK 160171B

77

Ved hjælp af en fasekomparator 52, der har sin ene indgang forbundet med udgangen på den programmerbare deler 51 og sin anden indgang forbundet med udgangen på overgangsdetektoren 53 føres disse impulser til 5 udgangen A eller udgangen R alt efter om de fasemæssigt ligger bagud eller forud i forhold til overgangene af en bestemt retning i det lokale taktsignal, hvilke overgange udgør periodebegyndelserne. De integrationsfiltre 54 og 55, der er koblet til kompa-10 ratorens udgange afgiver over en to-bit-udgang et signal til at opretholde eller rykke det lokale taktsignals fase frem eller tilbage, således at der opnås det samme antal impulser på de to udgange fra fasekompara-toren 52. Disse to filtre er f.eks. stokastisk rekur-15 sive filtre af typen "N before M" af den art, der er beskrevet af J.R. CESSNA i en artikel i IEEE Transactions on Communications bind COM. 20 nr. 2, april 72.By means of a phase comparator 52 having its one input connected to the output of the programmable parts 51 and its second input connected to the output of the transition detector 53, these pulses are fed to the output A or output R depending on whether they are phase lagged or preceded by relative to the transitions of a particular direction in the local clock signal, which transitions constitute the period beginnings. The integration filters 54 and 55 coupled to the comparator outputs output over a two-bit output a signal to maintain or move the phase of the local clock signal forward or backward so that the same number of pulses are obtained on the two outputs from the phase comparator 52. These two filters are e.g. stochastic recursive filters of type "N before M" of the kind described by J.R. CESSNA in an article in IEEE Transactions on Communications Volume COM. 20 No. 2, April 72.

Det ene filter har små værdier for N og M, hvorved der fås lille inerti, medens det andet filter har store 20 værdier N og M, der giver filteret stor inerti. Omskifteren 56 sætter det ene eller det andet integrationsfilter 54 henholdsvis 55 i drift. Som vist,kan omskifteren styres gennem en bistabil kipkobling 8 og en tæller 9. Når der ikke er nogen meddelelse på 25 koblingens udgang 2 holder modulationsdetektionssig-nalet D tælleren 9 på nul, og den bistabile kipkobling i en hvilestilling svarende til udvælgelsen af det integrationsfilter 54 eller 55, der har lille tidskonstant. Når der er en meddelelse på udgangen 2 fra 30 koblingen foretager tælleren 9 optælling af bittene i denne meddelelse, og når tælleren når op på sin maksimale kapacitet sender den en aktiveringsimpuls til den bistabile kipkobling 8, der skifter tilstand og indkobler det integrationsfilter, der har den store tids-35 konstant.One filter has small values for N and M, giving little inertia, while the other filter has large 20 values N and M giving the filter high inertia. The switch 56 activates one or the other integration filter 54 and 55 respectively. As shown, the switch can be controlled through a bistable tilt coupler 8 and a counter 9. When there is no message on the output of switch 2, the modulation detection signal D keeps the counter 9 at zero and the bistable tilt switch in a rest position corresponding to the selection of the integration filter. 54 or 55 that have little time constant. When there is a message at the output 2 of the coupling, the counter 9 counts the bits in this message and when the counter reaches its maximum capacity it sends an activation pulse to the bistable toggle switch 8 which changes state and engages the integration filter having the big time constant.

Kredsen 7 til hurtig fasegenvinding består iThe circuit for rapid phase recovery consists of

DK 160171 BDK 160171 B

8 hovedsagen af to dele, nemlig en kreds 71, der foretager detektering af en modulation, dvs. konstaterer tilstedeværelsen af en meddelelse i forsinkelseslinien 6 og den afgiver et modulationsdetekteringssignal D, 5 medens den anden kreds 72 tjener til genvinding af taktsignalet, dvs. foretager vurdering af den oprindelige fase af en meddelelse i forsinkelseslinien 6.8 is essentially two parts, namely a circuit 71 which detects a modulation, i.e. detects the presence of a message in the delay line 6 and it outputs a modulation detection signal D, 5 while the second circuit 72 serves to recover the clock signal, ie. evaluates the initial phase of a message in the delay line 6.

Modulationsdetekteringskredsen 71 kan udgøres af en auto-korrelator, der har to indgange forbundet med 10 enderne af forsinkelseslinien 6 og beregner autokorrelationsværdien mellem de to versioner af signalet i forsinkelseslinien 6, der indbyrdes er forskudt med en halv taktperiode, samt en komparator med dobbelt tærskel til frembringelse af et binært signal D, der 15 skifter til en tilstand, der svarer til en detektering af modulation, når den fra auto-korrelatoren afgivne auto-korrelationsværdi ligge over en nedre tærskel under hvilken man går ud fra, at det modtagne signal er støj, og under en Øvre tærskelværdi,over hvilken man 20 går ud fra, at der ikke er modtaget noget signal.The modulation detection circuit 71 may be constituted by an auto-correlator having two inputs connected to the 10 ends of the delay line 6 and calculating the autocorrelation value between the two versions of the signal in the delay line 6, which are offset by a half-stroke period, as well as a double threshold comparator to generating a binary signal D which switches to a state corresponding to a detection of modulation when the auto-correlation value emitted by the auto-correlator is above a lower threshold below which it is assumed that the received signal is noise , and below an Upper Threshold, above which one assumes that no signal has been received.

Taktsignalgenvindingskredsen 72, hvoraf to udførelsesformer skal beskrives, tilvejebringer på basis af signalerne ved enderne af forsinkelseslinien 6 et binært taktinitialiseringssignal , hvori overgangene 25 fra 0 til 1 nærmer sig begyndelserne af taktperioderne i en meddelelse i forsinkelseslinien 6, når denne meddelelse reelt er tilstede. Dette signal føres til en monostabil kipkobling 73, der udsender resæt-ningsimpulser til den programmerbare deler 51 i den 30 lokale taktgenerator 5, når den ikke spærres af signalet D fra modulationsdetekteringskredsen, dvs. udelukkende ved begyndelsen af hver detektering af modulation og for en enkelt udløsning.The clock signal recovery circuit 72, of which two embodiments are to be described, provides, on the basis of the signals at the ends of the delay line 6, a binary clock initialization signal, wherein the transitions 25 from 0 to 1 approach the beginning of the clock periods of a message in the delay line 6 when this message is actually present. This signal is fed to a monostable tilt coupling 73 which sends reset pulses to the programmable parts 51 of the local clock generator 5 when it is not blocked by the signal D from the modulation detection circuit, i.e. solely at the beginning of each detection of modulation and for a single trigger.

Den oprindelige vurdering af fasen af taktsigna-35 let i den modtagne meddelelse foregår ved at dele varigheden af en taktperiode i p successive tidsvinduer afThe initial assessment of the phase of the clock signal in the received message is done by dividing the duration of a clock period into successive time windows of

DK 160171BDK 160171B

9 ens varighed, og ved at gruppere disse vinduer i klasser i afhængighed af deres modulo p-rang i forhold til et arbitrært udgangstidspunkt, der vælges lokalt, og ved at bestemme den klasse vinduer, der oftest falder 5 sammen med den modtagne meddelelses overgange, der oplagres i forsinkelseslinien 6. Denne bestemmelse kan blot foregå ved i hvert tidsvindue, at tælle op det antal overgange i meddelelsen i forsinkelseslinien 6, der er faldet sammen med tidligere tidsvinduer af samme 10 modulo p-klasse, og ved at konstatere, om dette antal svarer til et maksimum.9, and by grouping these windows into classes according to their modulo p-rank relative to an arbitrarily selected local time, and by determining the class of windows that most often coincides with the transitions of the received message, which is stored in the delay line 6. This determination can only be made by counting in each time window, the number of transitions in the message in the delay line 6 that coincided with previous time windows of the same 10 modulo p class, and by ascertaining whether this number equals a maximum.

Fig. 2 viser en taktsignalgenvindingskreds, der virker på denne måde. Kredsen indbefatter to indgange 100 og 101, hvor den modtager de signaler, der er til-15 stede ved enderne af forsinkelseslinien 6, en taktstyreindgang 102, der modtager et rytmesignal , der definerer bredden af tidsvinduerne og har en frekvens på p gange frekvensen F og ved passende deling kan tilvejebringes ved hjælp af den lokale taktgenera-20 tors oscillator 50.FIG. 2 shows a clock signal recovery circuit acting in this way. The circuit includes two inputs 100 and 101 where it receives the signals present at the ends of the delay line 6, a clock control input 102 which receives a rhythm signal which defines the width of the time windows and has a frequency of p times the frequency F and by appropriate division can be provided by the local clock generator oscillator 50.

Enderne af forsinkelseslinien 6 er forbundet med to overgangsdetektorer 103 og 104, der afgiver impulser, der har samme bredde som et tidsvindue og er synkrone med overgangene.The ends of the delay line 6 are connected to two transition detectors 103 and 104 which emit pulses that are the same width as a time window and are synchronous to the transitions.

25 Et skifteregister 105, der har p trin og taktstyres af rytmesignalet virker som cirkulerende lager, der oplagrer det antal overgange i signalet i forsinkelseslinien 6, der er forekommet i de p klasser af tidsvinduer.A shift register 105, which is in step and clocked by the rhythm signal, acts as a circulating memory which stores the number of transitions in the signal in the delay line 6 that have occurred in the p classes of time windows.

30 En additionskreds 106, der danner sløjfe med skifteregisteret 105 tjener til ajourføring af de forskellige optællinger. Ud over en additionsindgang i sløjfen omfatter denne additionskreds en indgang til inkrementering med én enhed, hvilken indgang er forbun-35 det med udgangen på overgangsdetektoren 103, som er forbundet med indgangen til forsinkelseslinien 6, så-An additional circuit 106, which forms the loop register 105, serves to update the various counts. In addition to an addition input in the loop, this addition circuit includes one single-unit increment input which is connected to the output of the transition detector 103 which is connected to the input to the delay line 6,

DK 160171BDK 160171B

10 ledes at hver ny overgang tælles med, samt en indgang til dekrementering med én enhed, hvilken indgang er forbundet med udgangen på overgangsdetektoren 104, der er tilsluttet udgangen på forsinkelseslinien 6, hvor-5 ved hver overgang i det fra forsinkelseslinien 6 udgående signal tælles fra.10, each new transition is counted, and one unit decrement input, which is connected to the output of the transition detector 104 connected to the output of the delay line 6, at which each transition in the signal exiting the delay line 6 is counted from.

En detektor for relativt maksimum afgiver et binært signal på logisk niveau 1, hvis antallet af modulo p-klasser, som det igangværende tidsvindue hører 10 til, er større end det antal modulo p-klasser af tidsvinduer, der ligger umiddelbart foran eller følger u-middelbart efter det igangværende tidsvindue. Denne detektor består af to komparatorer 107 og 108, hver med to indgange, idet den ene komparator er tilsluttet 15 indgangen til og udgangen fra det første trin i skifteregisteret 105, medens den anden komparator er tilsluttet indgangen til skifteregisteret 5 og udgangen fra skifteregisterets sidste trin, samt en OG-port 109.A relative maximum detector outputs a binary signal at logic level 1 if the number of modulo p classes to which the current time window belongs 10 is greater than the number of modulo p classes of time windows immediately preceding or following the immediately after the current time window. This detector consists of two comparators 107 and 108, each with two inputs, one comparator being connected to the input and output of the first stage of the shift register 105, while the second comparator is connected to the input of the shift register 5 and the output of the last step of the shift register. , as well as an AND gate 109.

En tærskelkomparator 110, der er koblet til 20 udgangen på additionskredsen 106 spærrer udgangen fra detektoren ved hjælp af en OG-port 111 i det tilfælde, hvor det detekterede maksimum ikke overskrider en hvis tærskelværdi, som vælges i afhængighed af den støj, der kan påvirke den modtagne meddelelse.A threshold comparator 110 coupled to the output of the addition circuit 106 blocks the output of the detector by an AND gate 111 in the case where the detected maximum does not exceed a threshold value selected depending on the noise that may affect the message received.

25 Detekteringen af den modulo p-klasse af tids vinduer, der falder sammen med flertallet af overgange i en meddelelse i forsinkelseslinien 6, kan også foretages på basis af en bestemmelse blandt modulo p-klasserne af tidsvinduer af den klasse, hvori overgan-30 gene i meddelelsen fordeler sig mest retfærdigt til bege sider. Denne bestemmelse er mindre følsom overfor støjen, idet den giver mulighed for,for hver klasse at udnytte de fleste af overgangene i meddelelsen i forsinke Ise s linien 6 og ikke blot de overgange, der fal- 35 der sammen med tidsvinduerne i den pågældende klasse.The detection of the modulo p class of time windows that coincides with the majority of transitions in a message in the delay line 6 can also be made on the basis of a determination among the modulo p classes of time windows of the class in which the transitions occur the message is most fairly distributed to enthusiasts. This provision is less sensitive to noise, as it allows each class to take advantage of most of the transitions in the message in delay Ise s line 6 and not just the transitions that coincide with the time windows in that class.

Den har til gengæld den ulempe, at den fører til usik-On the other hand, it has the disadvantage that it leads to uncertain-

DK 160171BDK 160171B

11 kerhed,eftersom overgangene fordeler sig ligeligt til begge sider for den klasse tidsvinduer, med hvilke størstedelen af dem falder sammen, som til begge sider for den klasse tidsvinduer/ hvori størstedelen af dem 5 er faseforskudt med det halve af en taktperiode. Usikkerheden skal samtidigt fjernes, og dette sker ved at kontrollere, at tidsvinduerne i en klasse, der højst er -p/2 mindre end klassen af modulo p-rang, henholdsvis højst p/2 større end rangen af den klasse vinduer, 1 o hvori der er konstateret et minimum af fordelingsforskel, reelt ligger fasemæssigt forud eller bagud i forhold til størstedelen af overgangene i meddelelsen.11, since the transitions are distributed equally to both sides for the class of time windows with which the majority of them coincide, as to both sides for the class time window / in which the majority of them 5 are phase-shifted by half of a clock period. The uncertainty must be eliminated at the same time, and this is done by checking that the time windows in a class not more than -p / 2 smaller than the class of modulo p-rank, or at most p / 2 larger than the rank of the class windows, 1 o in which a minimum allocation difference has been found, which is actually phased ahead or backwards in relation to most of the transitions in the message.

Som tidligere nævnt, kan man i et tidsvindue få en optælling af de overgange i meddelelsen i forsinkel-15 seslinien 6, der falder sammen med tidligere tidsvinduer af samme modulo ρ-rang. Dette er også tilfældet for fordelingen af de overgange, der vedrører tidsvinduerne af samme modulo p-rang som det igangværende vindue, og som kan opnås ved hjælp af en rekurrent for-20 mel, hvori anvendes den fordeling, der blev kalkuleret for det tidsvindue, der kom før det igangværende vindue, sammen med optællingerne af de overgange, der er faldet sammen med visse klasser af modulo p-tidsvinduer.As previously mentioned, in a time window, one can get a count of the transitions in the message in delay line 6 that coincides with previous time windows of the same modulo ρ rank. This is also the case for the distribution of the transitions relating to the time windows of the same modulo p-rank as the current window, and which can be obtained by means of a competing formula using the distribution calculated for the time window. that came before the current window, along with the counts of the transitions that have coincided with certain classes of modulo p-time windows.

For at kunne vurdere fordelingen af overgangene 25 til begge sider af en klasse af tidsvinduer af rang i, skal man dele de tidsintervaller, der adskiller vinduerne i denne klasse af rang i i to lige store dele, hvoraf den ene ligger forud for hvert enkelt vindue og udgør det område, hvori overgangene betragtes som væran-30 de fasemæssigt forud,medens den anden følger efter hvert vindue og udgør det område, hvori overgangene betragtes som værende faseforsinkede, hvorefter man kalkulerer forskellen i fordelingen mellem de fasefremskudte overgange og de faseforsinkede overgange.In order to assess the distribution of transitions 25 to both sides of a class of rank windows of time, one must divide the time intervals separating the windows of this class of rank into two equal parts, one of which is preceded by each window and constitutes the area in which the transitions are considered to be phase-delayed, while the other follows each window and constitutes the area in which the transitions are considered to be phase-delayed, then calculating the difference in the distribution between the phase-advanced transitions and the phase-delayed transitions.

35 Fasefremskyndelsesområdet falder sammen med de tidsvinduer, hvis klasser har en modulo p-rang højst35 The phase advancement range coincides with the time windows whose classes have a modulo p rank at most

DK 160171BDK 160171B

12 p/2 mindre end rang i, medens faseforsinkelsesområdet falder sammen med de tidsvinduer, hvis faser har en modulo p-rang højst p/2 større end rang i. For at disse to områder kan have samme varighed, skal klassen af 5 vinduer med rang i være udelukket eller være indbefattet, alt efter om antallet p af tidsvinduer pr. taktperiode er ulige eller lige.12 p / 2 less than rank in, while the phase delay range coincides with the time windows whose phases have a modulo p rank at most p / 2 greater than rank in. For these two ranges to have the same duration, the class of 5 windows must be rank in be excluded or be included, depending on the number of time windows per clock period is odd or even.

Hvis det antages, at tallet p er ulige, er de overgange i den i transit i forsinkelseslinien 6 væren-10 de meddelelse fasemæssigt forud i forhold til tids-vinduerne i klassen af rang i de overgange, der er faldet sammen med de tidsvinduer, der hører til klasser af rang j, i henhold til betingelsen: i - k < j < i med k = i (p-1) , 15 medens de overgange, der er forsinkede i forhold til tidsvinduerne i klassen af rang i, er de overgange, der er faldet sammen med de tidsvinduer, der hører til klasser af rang j i henhold til betingelsen: 20 i<j^i + kAssuming that the number p is odd, the transitions in the message in transit in the delay line 6 are phased in advance of the time windows of the class of rank in the transitions that coincided with the time windows that belong to classes of rank j, according to the condition: i - k <j <i with k = i (p-1), while the transitions which are delayed relative to the time windows in the class of rank i are the transitions , which coincides with the time windows belonging to classes of rank ji according to the condition: 20 i <j ^ i + k

Fordelingsforskellen y^ mellem de fasefremskudte og faseforsinkede overgange i forhold til tidsvinduerne i klassen i, kan i afhængighed af antallet af overgange, der falder sammen med tidsvinduerne 25 af rang j udtrykkes på følgende måde: i - 1 i + k yi *3 xr 30 j = i-k r^TTT- hvilken relation kan ændres til: i-2 i+k-1 =/__ xj -/* xj+xi-i -xi-i-k+xrxi+k, 35 j = i-k-1 3 = i hvilken relation angiver fordelingsforskellen yi_1.The distribution difference y ^ between the phase-advanced and phase-delayed transitions relative to the time windows of class i, depending on the number of transitions coinciding with the time windows 25 of rank j, can be expressed as follows: i - 1 i + k yi * 3 xr 30 j = ik r ^ TTT- which relation can be changed to: i-2 i + k-1 = / __ xj - / * xj + xi-i -xi-i-k + xrxi + k, 35 j = ik-1 3 = in which relation denotes the distribution difference yi_1.

DK 160171BDK 160171B

1313

Idet det bemærkes, at indekserne skal ses i relation til modulo p: i-1-k = i+ k 5 opnår man rekursionsformlen: yi = yl-1* xi + xi-1 - 2 xi*k (1)Noting that the indices must be seen in relation to modulo p: i-1-k = i + k 5, we obtain the recursion formula: yi = yl-1 * xi + xi-1 - 2 xi * k (1)

Hvis antallet p af tidsvinduer pr. taktperiode er et helt lige tal, skal man medregne de vinduer i 10 klassen af rang i, der ligger i fasefremskyndelsesområdet eller faseforsinkelsesområdet, hvilket svarer til, at man for overgangene i meddelelsen som reference for fasefremskyndelse eller faseforsinkelse anvender begyndelserne eller enderne af vinduerne i klassen af rang 15 i. Hvis vinduerne af sidstnævnte klasse medregnes i faseforsinkelsesområdet ændres rekursionsformlen til:If the number p of time windows per. rate period is an even number, one must include the windows of the 10 class of rank i that are in the phase acceleration area or phase delay area, which corresponds to the use of the beginning or end of the windows in the class as a reference for phase acceleration or phase delay. of rank 15 in. If the windows of the latter class are included in the phase delay range, the recursion formula is changed to:

Yi = y±_.| + 2 xi_1 - 2 xi+k, med k' = 4 p, (2) medens den i modsat fald ændres til: 20 yi = yi-1 + 2 xi - 2 xi.k·· (3)Yi = y ± _. | + 2 xi_1 - 2 xi + k, with k '= 4 p, (2) while otherwise changing to: 20 yi = yi-1 + 2 xi - 2 xi.k ·· (3)

Fig. 3 viser et diagram over en taktsignalgenvin-dingskobling, der er egnet til de tilfælde, hvor der er et ulige antal p tidsvinduer pr. taktperiode, og hvori rekursionsformlen (1) anvendes.FIG. 3 shows a diagram of a clock signal recovery coupling suitable for cases where there is an odd number of time windows per minute. rate period and in which the recursion formula (1) is used.

25 På samme måde som i fig. 2 har denne kobling to indgange 100, 101, der er forbundet med enderne af forsinkelseslinien 6, samt en indgang 102, der får tilført et rytmesignal H^, der definerer bredden af tidsvinduerne,og hvis frekvens er et ulige multiplum p 30 af frekvensen F og opnås ved en passende deling af signalet fra oscillatoren 50. Den omfatter desuden et skifteregister 105 med p trin, der styres af rytmesignalet H.j og virker som cirkulationslager for antallene x^^ af overgange i meddelelsen i forsinkelses-35 linien 6, som er opstået i de forskellige modulo p-klasser af tidsvinduer, samt en adder 106, der er kob- 1425 In the same way as in FIG. 2, this coupling has two inputs 100, 101 connected to the ends of the delay line 6, and an input 102 which is supplied with a rhythm signal H 1 which defines the width of the time windows and whose frequency is an odd multiple of 30 of the frequency F and is obtained by an appropriate division of the signal from the oscillator 50. It further comprises a shift register 105 with p steps which is controlled by the rhythm signal H 1 and acts as a circulation memory for the numbers x 2 of transitions in the message in the delay line 6 which have occurred in the various modulo p classes of time windows, as well as an adder 106 which is coupled 14

DK 160171BDK 160171B

let tilbage til sig selv gennem et skifteregister 105 med p trin, der tjener til ajourføring af tælleresultaterne ved hjælp af en indgang, der inkrementerer med én enhed ad gangen, og som er forbundet med indgangen 5 til forsinkelseslinien 6 gennem en overgangsdetektor 103, og ved hjælp af en indgang til dekrementering med én enhed ad gangen ved hjælp af en overgangsdetektor 104, der er forbundet med udgangen på forsinkelseslinien 6.readily returns to itself through a p-step shift register 105 which serves to update the counter results by an input that increments one unit at a time and is connected to the input 5 of the delay line 6 through a transition detector 103, and at by means of an input for decrementing one unit at a time by means of a transition detector 104 connected to the output of the delay line 6.

10 Et andet skifteregister 120 med p trin, der styres af rytmesignalet / virker som cirkulationslager, der over én periode af taktsignalet holder de fordelingsforskelle y, der blev kalkuleret under de p forudgående tidsvinduer.Another shift register 120 with p steps controlled by the rhythm signal / acts as a circulation memory which holds over one period of the clock signal the distribution differences y that were calculated during the p preceding time windows.

15 En adder 121, der er koblet tilbage til sig selv gennem det første trin af skifteregisteret 120 kalkulerer fordelingsforskellen y^ mellem de overgange i den i transit gennem forsinkelseslinien 6 værende meddelelse, der fasemæssigt ligger forud henholdsvis 20 bagud i forhold til den klasse af tidsvinduer til hvilken det pågældende tidsvindue hører. Med henblik herpå udnytter den rekurrensformlen (1) og til det formål indbefatter den ud over sin additive tilbagekoblingsindgang, hvor den modtager den under det forudgående tids-25 vindue kalkulerede fordelingsforskel to andre additionsindgange, hvoraf den ene er forbundet med udgangen på det sidste trin i skifteregisteret 105, hvorfra den får det antal x.^ overgange, der falder sammen med den klasse vinduer af rang i modulo p, 30 til hvilken det pågældende tidsvindue hører, medens den anden indgang er forbundet med udgangen på det første trin i skifteregisteret 105, hvorfra den modtager det antal overgange, der falder sammen med klassen af vinduer af rang i-1 modulo p, samt en subtraktions-35 indgang, der gennem en 2-multiplikator 122 er forbundet med udgangen på trinnet i+ i (p+1) i skifte-An adder 121 coupled back to itself through the first step of the switch register 120 calculates the distribution difference γ mellem between the transitions in the message passing through the delay line 6 which are phased ahead and 20, respectively, relative to the class of time windows. to which that time window belongs. To this end, it utilizes the recurrence formula (1) and for that purpose includes, in addition to its additive feedback input, where it receives the distribution difference calculated during the preceding time window two other addition inputs, one of which is connected to the output of the last step of the the switch register 105, from which it obtains the number of x. ^ transitions that coincides with the class windows of rank in modulo p, 30 to which that time window belongs, while the second input is associated with the output of the first step of the switch register 105; from which it receives the number of transitions coinciding with the class of windows of rank i-1 modulo p, as well as a subtraction input connected through the output of step i + i (p + 1) through a 2 multiplier 122 switch-

DK 160171BDK 160171B

15 registeret 105, hvorfra den modtager tælleresultatet X-j/Hp+D -15 register 105 from which it receives the counter result X-j / Hp + D -

Den modulo p-klasse af tidsvinduer i forhold til hvilken overgangene i meddelelsen i forsinkelseslinien 5 6 fordeler sig mest ligeligt til begge sider,er den klasse, for hvilken den numeriske værdi af fordelingsforskellen y er minimal. Søgningen efter dette minimum sker ved hjælp af to kredse 123 og 124, der tilvejebringer den numeriske værdi, samt en komparator 10 125, hvis to indgange er koblet til nævnte kredse.The modulo p class of time windows relative to which the transitions in the message in the delay line 56 are distributed most equally to both sides is the class for which the numerical value of the distribution difference y is minimal. The search for this minimum is done by means of two circuits 123 and 124 which provide the numerical value, as well as a comparator 10 125 whose two inputs are coupled to said circuits.

Værdifrembringelseskredsen 123 er forbundet med udgangen fra adderen 121, hvorfra den får tilført den fordelingsforskel y^, der gælder for modulo p-klassen af det igangværende vindue, medens værdifrexnbringelses-15 kredsen 124 er forbundet med udgangen på det sidste trin i skifteregisteret 120, hvorfra den får tilført fordelingsforskellen yi+1 for den modulo p-klasse af tidsvindue, der følger efter det igangværende vindue. Komparatoren 125 afgiver et binært signal, der har 20 logisk tilstand 1 sålænge den numeriske værdi af fordelingsforskellen yi for modulo p-klassen af det i-gangværende vindue er mindre end fordelingsforskelien y^+^ for modulo p-klassen af det tidsvindue, der følger efter.The value generation circuit 123 is connected to the output of the adder 121 from which it is applied to the distribution difference y1 which applies to the modulo p class of the current window, while the value generation circuit 124 is connected to the output of the last step of the switch register 120, from which it is added to the distribution difference yi + 1 for the modulo p class of time window that follows the current window. The comparator 125 outputs a binary signal that has logic state 1 as long as the numerical value of the distribution difference yi for the modulo p class of the current window is less than the distribution difference y + + + for the modulo p class of the time window that follows. after.

25 To tærskelkoblere 126 og 127 styrer udgangs signalet fra komparatoren 125 gennem en OG-port 128, der afgiver et taktinitialiseringssignal . Deres opgave er netop at fjerne den før omtalte usikkerhed. Komparatoren 126 er tilsluttet udgangen på det trin 30 af skifteregisteret 120, hvis rang er i nærheden af en 1/4 (p-1) og den afgiver et binært signal med logisk tilstand 1, hvis den fra skifteregisteret 120 modtager en negativ fordelingsforskel y. Den anden tærs-kelkomparator 127 er forbundet med udgangen på et 35 trin i skifteregisteret 120, hvis rang er i nærheden af 1/4 (3 p-1) og den afgiver et binært signal medTwo threshold couplers 126 and 127 control the output signal of comparator 125 through an AND gate 128 which delivers a clock initialization signal. Their task is precisely to remove the previously mentioned uncertainty. The comparator 126 is connected to the output of the step 30 of the shift register 120, whose rank is in the vicinity of a 1/4 (p-1) and it outputs a binary signal with logic state 1 if it receives from the shift register 120 a negative distribution difference y. The second threshold comparator 127 is connected to the output of a 35 step of the shift register 120, the rank of which is in the vicinity of 1/4 (3 p-1) and outputs a binary signal of

DK 160171BDK 160171B

16 logisk tilstand 1, hvis den fra skifteregisteret 120 modtager en positiv fordelingsforskel y.16 logic state 1 if it receives from the shift register 120 a positive distribution difference y.

Forskellen y ved den udgang fra skifteregisteret 120, der har rang i nærheden af 1/4 (p-1) har til-5 knytning til en klasse tidsvinduer, hvis rang er ca.The difference y at the output of the shift register 120 having rank near 1/4 (p-1) has -5 association with a class of time windows whose rank is approx.

1/4 (p-1) mindre end rang i af det igangværende tidsvindue, og som derfor ligger ca, 1/4 taktperiode forud for tidsvinduerne af rang i. Når den numeriske værdi af fordelingsforskellen y^ passerer gennem et mini-10 mum for tidsvinduerne af rang i, er den nødvendigvis negativ, hvis disse vinduer falder sammen med størstedelen af overgangene i meddelelsen og den er nødvendigvis positiv, hvis disse vinduer er forskudt en halv taktperiode i forhold til størstedelen af overgangene i 15 meddelelsen. Denne situation udnyttes af tærskelkompara-toren 126 for at undertrykke detekteringen af det absolutte minimum, der svarer til labil ligevægt.1/4 (p-1) less than the rank in of the current time window, which is therefore approximately 1/4 clock period ahead of the time windows of the rank in. When the numerical value of the distribution difference y ^ passes through a mini-10 mum of the time windows of rank i, it is necessarily negative if these windows coincide with most of the transitions in the message and it is necessarily positive if these windows are offset by half a clock period compared to most of the transitions in the message. This situation is exploited by the threshold comparator 126 to suppress the detection of the absolute minimum corresponding to labile equilibrium.

Fordelingsforskellen y på den udgang af skifteregisteret 120, hvis rang er i nærheden af 1/4 (3 p -20 1), har tilknytning til en klasse tidsvinduer, hvis rang er ca. 1/4 (3 p-1) mindre end rang i af det igangværende tidsvindue, eller, ud fra definitionen af modulo p-rang, har ca. 1/4 (p-1) større rang end rang i af det igangværende tidsvindue og ligger derfor ca. 1/4 25 taktperiode efter tidsvinduerne af rang i. Når den numeriske værdi af fordelingsforskellen y^ passerer gennem et minimum for tidsvinduerne af rang i, er den nødvendigvis positiv, hvis disse vinduer er forskudt 1/2 taktperiode i forhold til størstedelen af overgan-30 gene i meddelelsen. Denne tilstand udnyttes af tærskel-koraparatoren 127 for at udelukke detekteringen af det absolutte minimum, der svarer til labil ligevægt. Ved hjælp af tærskelkomparatorerne 126 og 127 får taktinitialiseringssignalet således overgange fra 35 0 til 1, der kun forekommer i de tidsvinduer, der falder sammen med begyndelserne af taktperioderne i meddelelser-The distribution difference y on the output of the shift register 120, whose rank is in the vicinity of 1/4 (3 p -20 1), is associated with a class of time windows whose rank is approx. 1/4 (3 p-1) less than rank in of the current time window, or, from the definition of modulo p rank, has approx. 1/4 (p-1) greater rank than rank i of the current time window and is therefore approx. When the numeric value of the distribution difference y ^ passes through a minimum for the time windows of rank i, it is necessarily positive if these windows are offset 1/2 rate period relative to most of the transition time. 30 genes in the message. This condition is utilized by the threshold correlator 127 to exclude the detection of the absolute minimum corresponding to labile equilibrium. Thus, by means of the threshold comparators 126 and 127, the clock initialization signal receives transitions from 35 0 to 1 which occur only in the time windows coinciding with the beginnings of the clock periods in messages.

Claims

A coupling for recovering the beat signal from an NRZ-encoded binary message, and of the kind comprising 15 rhythmic means for generating a rhythm signal () having a frequency equal to a multiple p of the frequency (F) of the binary message clock signal defining successive time windows of equal duration, counting means (105, 106) operating on the signal transitions received over more than 20 clock periods, and maximum detecting means (107, 108) operating on the obtained count values, characterized by the counting means (105, 106) are arranged to group the time windows into classes depending on their modulo p-rank relative to an arbitrary starting time, and to determine within each time window the number of (X ^) transitions that have coincided with previous time windows of the same class (i) modulo p as the current time window, and that the maximum detection means are arranged to output a clock initialization signal (H is part of the modulo p class having the maximum count value, the coupling further having a local clock generator (5) having the received message clock frequency and which is synchronized at the beginning of the recovery procedure by the maximum detecting means. 18 DK 160171B

2. Coupling to recover the beat signal from an NRZ-encoded binary message, and of the nature comprising rhythmic means for generating a rhythm signal (H1) F having a frequency equal to a multiple p of the frequency (F) of the binary message clock signal defining successive time windows of equal duration, counting means (105, 106, 120, 121, 122) operating on the signal transitions received over multiple clock periods, and minimum detection means (123, 124, 125) operating 10 on the resulting distribution differences, characterized in that the counting means (105, 106, 120, 121, 122. are arranged to group the time windows into classes depending on their modulo p-rank relative to an arbitrary starting time and to within each time-15 window to determine the distribution difference () between the transitions ahead and the transitions that lag behind the time windows of the same class (i) as the current window that the minimum detection means e (123, 124, 125) are adapted to output a tactic initialization signal in the time windows belonging to the class modulo p which exhibits the lowest distribution difference, the coupling further having verification means (126, 127) for blocking the minimum detection means. , when the distribution difference for a class with modulo p rank 25 is maximum | p less respectively \ p greater than the rank of the class of time windows for which a detected minimum in the distribution difference does not allow for the benefit of the delayed or delayed transitions, and a local clock generator (5) having the message frequency of the received message, and which is synchronized at the beginning of the synchronization procedure by the minimum detection means (123, 124, 125).

Coupling according to claim 2, characterized in that the counting means include: 35. counting means (105, 106) which are activated by the signal transitions received over several clock periods and group the time windows into classes depending on their modulo p-rank relative to an arbitrary exit time and within each time window determine the number of (x ^) transitions coinciding with previous time windows of the same class (i) as the current window, and calculating means (120, 121, 122) that for each current time window provides the value of the distribution difference (y ^) for the class (i) of the current window by recursive calculation on the basis of the distribution difference (y ^ _ j) calculated during the preceding time window and of the most recent results previously was indicated by the count means (105, 106) for the counts of different classes of time windows.

The coupling according to claim 3, wherein p is an odd number, characterized in that the calculating means (120, 121, 122) in each time window give the value of the distribution difference y ^ based on the recursion formula * i = yi-1 + Xi + Xi-1 "2 Xi + k 20 where k = in thaw" 1) ·

Coupling according to claim 1, characterized in that the counting means include: a shift register (105) with p steps controlled by the rhythm signal (H 1), and 25. An additional circuit (106) with addition input easily switches its output through said switch register (105) and having an input for incrementation with "1", which input is associated with the output of a first transition detector (103) receiving the message without delay, and with an input for decrementing with "1", which input is connected to the output of another transition detector (104) receiving the message at a given delay.

The coupling according to claim 3, wherein p is an odd number, characterized in that the counting means include: * a counting means with a shift register (105) with p steps controlled by the rhythm signal (H),. an addition circuit (106) with addition input 5 coupled to its output through said switch register (105) and having an input for incrementation with "1", which input is connected to the output of a first transition detector (103) receiving the message without delay, and having a 10 "" decrement input which is connected to the output of another transition detector (104) receiving the message at a given delay and calculating means by a doubler (122) associated with the output at the step of rank i (p + 1) of the shift register (105) in the counting means; • a shift register (120) controlled by the rhythm signal (Hj); and • an addition circuit with addition input coupled to its output through the first step of the counting means. switch register (105), and with a second and a third input connected to the output of the first and the output of the last step of the switch register (105), and having a subtraction input connected to the output of the doubler 25 (122).

Coupling according to claim 6, characterized in that the minimum detecting means comprise: two between circuits (123, 124) for generating numerical value between the input of the switch register (120) and the output of the register means for generating numerical value, and a comparator ( 125) with two inputs connected to said numeric value circuits to provide a clock initialization signal each time the numerical input value of the shift register (120) is less than the value of the output of the shift register p step. DK 160171B 21

Coupling according to claim 6, wherein p is an odd number immediately following a number divisible by four, characterized in that the verification means comprise: 5 a current threshold comparator (126) connected to the output of the step of rank 1 / 4 (p-1) in the switching register (120) of the calculating means and blocking the minimum detection means as long as the signal to the comparator is positive, and 10. a current threshold comparator (127) associated with the output at the step of rank 1/4 (3p-1) in switching register of the averaging means and blocking the minimum detection means as long as the signal for this comparator is negative.