DE2043144A1 - Digital transmission path with regenerative amplifiers - Google Patents

Description

2043H4

Patent attorneys Dlpl.-Ing. R. BEETZ sen.

DIpI-In ₉₁ K ₁ LAMPMECHT ΐ25-ιβ.θβ7Ρ August 31, 1970

Dr.-lng. R. BEET 2 Jr 8 Manchen 22, Steinsdorfetr. 10

KRONE limited partnership, Berlin

Digital transmission line with regenerative amplifiers

The invention relates to a digital transmission link for the transmission of digital signals, with regenerative processors arranged between the individual transmission link sections for the regeneration of the digital signals distorted and noisy by the attenuation distortion of the transmission link section immediately preceding in the transmission direction, the regeneration boosters each having an equalizer for the incoming digital signals have whose output a from the single pulses of the output signal of the equalizer when exceeding a positive or negative amplitude threshold * pulses corresponding to the individual pulses of the undistorted digital signal generating Amplitudenregenerierkreis and to a

125-x1560-Hd-r (7)

2098U / 1287

~ ² ~ 2043H4

cancel the phase errors of the output signal of the equalizer Zeitregenerierkreis is connected, with de circles a regenerator to deliver the regenerated Control digital signal.

The digital signals are usually made up of analog signals such as voice signals, measured value signals or the like obtained because the transmission of digital signals basically more interference-free than from analog signals vorfc can be taken «To do this, analog signals have to be converted into digital form. The analog signal first sampled according to the sampling theorem

The sampling theorem says that a band-limited signal function whose Fourier spectrum U (f) is zero ^ is exactly determined for all frequencies f ^ B by its discrete sampling values at a distance of T ^ 1/2 B ₀ The sampling theorem is of fundamental importance for transmission technology, since, according to his statement, it is sufficient if the amplitude of a band-limited signal is sampled and transmitted at equidistant time intervals U »

ψ Ver

different methods or types of modulation are known.

A first method is pulse amplitude modulation (PAM), in which the amplitudes of pulses are made equal to the amplitudes of the sampled values. A second The procedure is the pulse phase modulation (PPM), in which the time position (phase) of the pulses is proportional to the Size of the samples changed.

A third possibility is the modulation of the im-

2096U / 1267

pulse duration, pulse duration modulation (PDM) · The duration the pulses are modulated proportionally to the amplitude of the samples.

Pulse code modulation (PCM) has been widespread (cf. BM Oliver, J, R. Pierce, C, E ₀ Shannon, The Philosophy of PCM, Proc. IRE, November 1948).

In this modulation method, an analog signal is first subjected to pulse amplitude modulation.

The sampled values of the analog signal obtained in this way are then quantized and fed one after the other to an encoder in the form of an analog-digital converter. The encoder assigns a code word to each quantized sample value, which code word consists of a sequence of a certain number of code symbols which, for. B. consists of two different voltage or current states (voltages + U and -U) usually with "O" and "1 ^H" individual pulses, which together form the digital signal (see Fig. Ia) ₀ Since the code symbols are only two different This type of coding is called binary code, but other, related codes for digital signals will be discussed later.

The digital signal is then fed into a transmitter and transmitted to a receiver via a transmission link. The transmission path may be, for ₀ B ₀ to be a line in an electrical cable, a radio link or the like.

After passing through the transmission link, the Receiver converts received digital signal into an analog signal

20431U

decoded, which, in the case of ideal transmission, is sent to the sender Analog signal.

If you have digital transmission routes in the form of z. B. wants to transmit longer lines (10 km and longer) very fast digital signals that z. B ₀ originate from pulse code modulation (PCM) devices or other data sources and have a single pulse or symbol repetition frequency (bit frequency for binary code), also called clock frequency in the following, of a few MHz, regenerative amplifiers must be provided in the transmission links, which are in the meantime on the transmission link regenerate distorted and disturbed signals in terms of amplitude and phase, as will be explained in more detail (principle of the regenerative path).

1c shows the basic circuit diagram of such transmission links known per se. A transmission line 2, which forms a first transmission link, extends from a first terminal 1 to a second terminal 3, from which a receiving line h , which forms a second transmission link, runs back to the first terminal 1. The transmitting line 2 and the receiving line k are designed as double lines, which is indicated in FIG. 1c by two dashes on the lines. Regenerative boosters 5 are interposed in the transmitting line 2 and the receiving line k at relatively short intervals (for example between 1 and 3 km); which divide the lines or transmission lines 2 and k into line or transmission line sections.

2098U / 1 287

2043HA

The Regeneratiwerstärker 5 are used on the in each case preceding transmission link sections in the transmission direction to regenerate distorted and disturbed digital signals with regard to amplitude and phase (in to bring the status at the output of the associated sending station) and then to the next transmission link section to be handed in

Theoretically, you can provide any length of transmission links in this way without the interference always grow. This is the decisive advantage of transmission systems for the transmission of digital signals compared to transmission systems for the transmission of To see analog signals. Because of an amplitude threshold (AS in Fig. 1a) the regenerative boosters must, especially in the case of binary code, the "1" individual pulses of the digital signal when the regenerative boosters are transmitted Regeneration intensifiers remain above this amplitude threshold, since otherwise it regenerates incorrectly as "O" pulses will.

The regenerative amplifiers 5 normally draw their feed or supply power via the transmission line, that is to say via the same line over which the digital signal also runs. The feed power can z _o B ₀ via a phantom circuit (vglo Steinbuch-Rupprecht, Telekommunikation, 1967, page 261) »

In some cases it is desirable for security reasons to limit the power to be transmitted. Often it is not with such a transmission path possible, the supply voltage or the supply

2098U / I287

disable current when repairs have to be made ₀ Therefore, the maximum occurring on the transmission line voltage to ground z _o to B ₀ not exceed 60 V and the maximum current occurring z _s B ₀ 50 mA.

In general, the supply current for the regenerative boosters is fed into the phantom circuit from the terminals as a constant current of 50 mA.If you take into account that the maximum voltage does not exceed 60 V and you use +60 V at one terminal and +60 V at the other terminal -60 V feeds in, a total voltage of 120 V can be expected. In addition, one must take into account ₉ that due to the ohm ¹ resistance of the transmission line, about 3 V drop in each line section, so not all of the voltage available can be used for the regenerative amplifier »

It has therefore proven to be useful to branch off each regenerative amplifier over 5 V with the help of a Zener diode. This ensures that minde- least ten W amplifiers may be fed to a track and each amplifier 5 V and 50 mA, so 250 mW has as a supply power is available, with which one must deal seen very economical.

As will be explained in a regeneration booster many analog arithmetic operations and filter operations are carried out, for which one can find in the Electronics normally used operational amplifiers, which, however, have a large voltage and current requirement

2098U / 1 287

2CU31U

show o Because of the low power available But in regeneration you are forced to use discrete components such as transistors, diodes, field effect transistors and to use transformers. That is why the well-known regeneration boosters are very large and unwieldy.

This is very disadvantageous because it can happen that hundreds of Regeneratiwers strengthen in a cable duct must be housed where space is very limited. If the regeneration booster housing too still have to be connected to a cable monitored with compressed gas, grows with the size of the regenerative boosters and with the size of the regenerative booster housing, the problem is that these housings are absolutely watertight and gas-tight to get.

A section of any of the transmission links from FIG. 1 with the block diagram of the η-th regenerative amplifier 5 in the transmission direction and a part of the (n + i) -th regenerative amplifier 5 for the digital signal encoded in binary code can be seen in FIG. 2a, η can have any positive integer value. The structure of the regenerative amplifier shown in FIG. 2a is essentially known per se (cf. van der Houwen, Het PTT-Bedrijf, XVI _; No. 2, March 1969, p. 81 ff, in particular FIGS. 2 and k) .

At various points the regeneration booster Signals occurring from Fig. 2a are shown in Fig. 3, in which with I the output from the sending terminal, distortion and attenuation-free digital signal is designated in binary code.

2098U / 1287

- ⁸ "20431U

The digital signal to be regenerated is fed from the (ni) th regenerative amplifier (not shown) via an output transformer into the η th line section and from there into an input transformer 10 of the η th regenerative amplifier, which is connected to an equalizer 11. In the equalizer 11 of the η-th regenerative amplifier, a signal II (cf. FIG. 3) is processed in such a way that the attenuation and the attenuation distortion of the n-th transmission link section are reversed. At the equalizer output 11 a signal III occurs (see. "Fig. 3), which is fed to serving as a summing amplifier precursor 122, and a downstream decision 123 of a Amplitudenregenerierkreises 12 _e The discriminator 123 is a threshold detector is exceeded its threshold A "1" level is generated by the input signal at its output, but otherwise an "O" level. In this way, the digital signal is regenerated in terms of amplitude, which is why stage 12 of the regeneration amplifier is called an amplitude regeneration circuit (cf. FIG. ₀ 2a),

The output signal of the decision maker 123 is pending no regenerator 16, which can be a bistable multivibrator and only takes over the information when from a pulse shaper 1> a time window signal IX comes, whose Individual impulses for temporary regeneration of the digital signal at axact intervals according to its individual impulse duration or clock period T must follow one another.

The transformer coupling (instead of a galvanic) of the digital signal from the transmitter or from one of the regenerative amplifiers to the individual sections the transmission line and from it to the regenerative

2098U / 1787

" ⁹ " 20431U

amplifier through the aforementioned transformer has the following advantages: The individual sections of the transmission path can be floating and adapted to the regenerative amplifier resistance, and the frequency zero can be reserved for the aforementioned remote feed of the regenerative amplifier. However, as a result of the transformer coupling, the amplitude, in particular of the "1" individual pulses of the digital signal free of distortion and attenuation, falls somewhat over time (cf. dashed lines in FIG. 1a). If several "1" single pulses immediately, i.e. h "without intervening" O "single pulses, it can happen that the amplitude of the last" 1 "single pulse or pulses has fallen below the amplitude threshold value of the regeneration boosters, so that these" 1 "single pulses are classified as" O "- Individual pulses are regenerated ₀ This effect is known as zero line migration from telegraphy.The zero line migration can also be understood to mean that the digital signal present in binary code has a high direct current content, i.e. a strong Fourier component of the frequency zero, so that during a transmission of digital signals without the possibility of transmission for the frequency zero, the considerable information relating to the component with the frequency zero is lost, _etc.

The direct current component is therefore still missing in signal III and the low-frequency frequency components of the digital signal encoded in binary code, which are in the output transformer of the (n-i) -th Rgeneratiwerstärkers and in the input transformer 10 of the η-th regenerative booster as a result whose high-pass behavior has been lost. In order to feed these frequency components to signal III, © in DC recovery filter 124 provided that part

2098K / 1287

20431U

of the amplitude regeneration circuit 12 and from the regenerator 16 to the preliminary stage 122. The dimensioning of the filter 124 is difficult because several time constants have to be taken into account, for example that of the input transformer 10 and that of the output transformer 18 of the preceding regenerative amplifier. Van der Houwen (a, a _o 0.) therefore suggests dimensioning the time constants of these transformers very differently, so that the larger one can be neglected. This procedure would be correct if the transformers were to work under exact conditions.But this is not the case, especially at the low frequencies for which these considerations (see above) apply, if transmission is to be carried out over a transmission link section in the form of a symmetrical cable , because its wave resistance is difficult to define and complex at low frequencies.

For explanation, FIG. 2b shows the equivalent circuit diagram for a transmission system consisting of the transmission output stage 17 (generator with internal resistance R 1), the output transformer 18, the transmission link section (with characteristic impedance Z _T ). the input transformer 10 and an output

“Lj
resistance R shown »It can be seen that both over-

Gt

carriers 18 and 10 from one side with the wave resistance Z are wired, which is bad at low frequencies definable and not real. Because the time constants of the transformers 18 and 10 are now derived from the formulas

R. Z-. RoZ

™ i L 1 "a L 1

¹ Ie ^- R ₄₊ Z. L ¹ 10 R ₊ Z- L

1 ly U L

calculate, i.e. are dependent on ZL, these cannot be specified precisely for low frequencies.

2098 U / 1207

20431U

The time window signal IX serves - as mentioned - for the temporal regeneration of the digital signal, which is why the the partial circuit of the regenerative amplifier generating the time window signal IX, consisting of a clock recovery device 14 and the pulse shaper 17 »with time regeneration circuit (Fig. 2a) is designated.

The time window signal IX is obtained as follows:

Fig. Ka shows a sinusoidal approximation of the output signal of the equalizer III 11, the undistorted and undamped rectangular, in binary coded digital signal (binary signal called short) of FIG. Kh corresponds. The power density S (x) as a function of the normalized frequency χ = f · T (T = single pulse, or clock period) for the sinusoidal approximate binary signal (ILL) is shown in Fig. Kc It is assumed that all statistically possible sequences of ^M 1 * And ⁿ 0 ^w single pulses occur in the binary signal (ill in Fig. 3).

From Fig. 4c it can be seen that the power density for the single pulse repetition frequency or bit frequency 1 / T disappears. So you cannot get the single pulse train or clock frequency directly from the binary signal (ill) Filter out f = 1 / T.

For this reason, the binary signal a transformation is subjected, so that the power spectrum of the transformed signal is a maximum of half the single-pulse repetition frequency (see. Fig. Kc) by the binary signal (ILL) is differentiated by a differentiator 131 in a control device 13 . These differences

2098U / 1287

2G43U4

In terms of information theory, tiation corresponds to a code conversion, since the transformed signal represents the information contained in the binary signal (ill) in a different form. The differentiated binary signal (ill) is equal to the signal IV in FIG. J _a. In the control device 13, a non-linear filter 135 is connected to the differentiator 131, which derives trigger pulses from the signal IV, which the signal V (cf. _o FIG. 3) form and the component of frequency or Einzelimpulsfolge- included because enters a w frequency doubling by the nonlinearity of the filter 135th

The trigger pulses of the trigger signal V are, however, subject to phase errors caused by interference, attenuation distortion or causes non-ideal equalization could be. These phase errors are determined by the clock recovery device 14, as will be explained shortly.

A Schmitt trigger 151t, which generates the signal VIII from the signal VII, is connected to the clock generation device \ h , and a differentiating element 152 is connected to this, which generates the time window signal IX.

The clock recovery device \ h is designed in such a way that it can also overcome long clock pauses which arise from the fact that long "1" single-pulse sequences or "0" single-pulse sequences are contained in the digital signal. For this purpose, the clock recovery device Ik has a high-quality quartz filter 142 which is matched to the clock frequency, which generates weakly damped oscillations and contains an operational amplifier with a quartz as input resistor

2098U / 17 87

Clock information, i.e. very many "1" - "0" - "1" -WeChSeI shows, does not need to use the entire energy of the impulses to become.

In this context it is already known (cf. van der Houwen, op. Cit. 0) to provide such a control, that the width of the trigger pulses V for the timing regeneration circuit is controlled in such a way that the trigger pulses V are very narrow when there is a lot in the digital signal Clock information is included, and very wide if a long clock pause bridged between "1" individual pulses became. Such a control within the non-linear filter 135 (cf. FIG. 2a) should be based on FIG. 2c and 2d will be explained.

The digital signal, which is processed by the differentiator 131 so that the maximum of its power spectrum is at half the single pulse repetition frequency, reaches the base of a transistor T. This digital signal, which is shown at the top in FIG. 2d (£ IV in FIG. 3) would occur in the form shown at the emitter of the transistor T if the capacitor C "were not to be charged. As a result of the capacitor C_, the signal profile at the emitter of the transistor T is approximately according to the dashed curve in FIG. 2d. That is, the transistor T can only carry current at very short time intervals At, so that the narrow trigger pulses (SV in FIG. 3) for the subsequent crystal filter 1 ^ 2, shown at the bottom in FIG. 2d, arise at the collector of the transistor T. These impulses therefore have the duration / \. t.

As can be seen from the course of the dashed curve in FIG. 2d above, however, this width is At thereof

2098U / 1? 87

depends on how big the distance to the previous single pulse is. D ₀ h. if this distance is small, the nudge pulses will also be very narrow and contain little energy, and if the distance is large, the first nudge pulse will be very wide and accordingly contain a great deal of energy. This realizes a control which somewhat compensates for the different clock content ("0" - "1" - "0" -WeChSeI) of the digital signal.

^ A very fine dosage that precisely matches the toasting energy requirement the clock recovery device 14 is adapted, but is not possible because of its output signal VII is not detected and therefore, to be on the safe side, you have to work with a little excess energy. aside from that this control has the disadvantage that after a long pause only the first trigger pulse is widened and the following, if there is no large pause, remain narrow.

When a long clock pause has been bridged, the output amplitude of the quartz filter 142 will have already decayed a great deal. Since the quartz filter 142 works with a ψ high quality (about 1000), a more energetic trigger pulse is not sufficient to bring the output amplitude of the quartz filter 1 ^ 2 back to a high value.

Therefore, the clock recovery device ^ k must be controlled with even greater excess energy (relatively broad trigger pulses) and then also takes over many phase errors from the digital signal.

At the output of the clock generation device 14, which is the crystal filter 142 in the known circuit, should

2098U / 1287

a sinusoidal signal occur so that the subsequent Schmitt trigger 151 becomes a square-wave signal without phase errors can form. (Phase errors arise from variable temporal position of the edges of the square wave signal). However, the quartz filter 142 is not an ideal resonance filter for the fundamental frequency of the quartz · The reasons for this are to be found in the properties of the quartz. The quartz can namely not only excited to its fundamental frequency, but also to harmonics of the fundamental frequency. Even stronger disturbs the existing parallel capacitance of the quartz, the differentiating properties of the quartz filter 1 ^ 2 there (an operational amplifier with a capacitance as an input resistance acts as a differentiator), which is why at the output of the clock recovery device 14 through Peak distorted sine waves appear. The subsequent Schmitt trigger 151 can be derived from the distorted sinusoidal oscillations do not make a perfect, symmetrical square wave signal, so that the edges of the square wave signal Occur offset in time, so a phase error occurs.

The output signal ΧΧΣΙ of the regenerator 16 is therefore regenerated only within the limits indicated above in terms of amplitude and time and is via a stage 17 »die from a transmitter and a connected low-pass filter exists in order to attenuate the disruptive spectral components above half the single pulse repetition frequency, to one Output transmitter 18, at the output of which a signal XIV occurs, (in this context it should be mentioned that, according to van der Houwen, the equalizer 11 is not located immediately after the input transformer 10, but is immediately sounded out located in front of the output transformer 18 of the regenerative amplifier.)

209SU / 1287

" ¹⁶ " 2043 HA

The equalizer 11 of each regenerative amplifier serves, together with the transmission low-pass filter of stage 17 of the in the transmission direction preceding regenerative amplifier and the transmission link section between the Output transformer 18 of the previous regeneration amplifier and its own input transformer 10, one for error-free amplitude regeneration of the digital signal to form the most favorable transfer function possible. Therefore, the combination of the transmission low-pass filter of the nth ^ Low-pass filter of the (n + i) -th transmission path ^ Section and equalizer of the (n + · i) -th regenerative amplifier denoted by the nth transmission unit, as is also shown in FIG. 2a.

The choice of the optimal course of the transfer function results from considering the types of Sources of interference. The following sources of interference must be taken into account:

1. Interference caused by similar digital signals in an adjacent similar transmission path, e.g. B. in a neighboring line in a cable, indicated by the so-called near-end crosstalk coupling (by dashed line arrows L · in Fig. Is). The high-frequency spectral components of the digital signal are the most disturbing because the (predominantly capacitive) coupling between the two similar transmission paths is strongest for high frequencies.

From this it follows in a manner known per se (cf. van der Houwen, Het PTT-BedriJf, XVI, No. k, February 1970,

200814/1287

" ¹⁷ " 2Q43U4

Page 212 ff ₀ ) that the transmission unit defined above is expediently a low-pass filter, the upper limit frequency of which should be shifted downwards with increasing transmission link length External interference becomes less, as van der Houwen has shown numerically. If one regards only near-end crosstalk as interference, there is an optimum of this overlap (cf. ₀ Van der Houwen, aa 0., pages 213 and 214) »

This overlap is partly predictable or -bestimmbar (see FIG. Van der Houwen, a. A ₀ 0., pages 213 »2i4) o What is meant by" non-predictable "and under" predictable "percentage of overlap _s is based on Fig 5 can be seen.

FIG. 5b shows the case of a single "1" single pulse which has been obtained by the transmission unit acting as a low-pass filter from a single rectangular ^M 1 ^W single pulse of the digital signal according to FIG. 5a. This means that the simple case is assumed that in the digital signal only "O" single pulses occur before and after the "1" single pulse.

The threshold in FIGS. 5a and 5b means the araplitude threshold of the decision maker in the amplitude regeneration circuit 12. With t and correspondingly t ₁ and t ₁ , times are indicated at which very narrow individual pulses of the time window signal IX are delivered at the time interval T from the pulse shaper 15 of the regeneration circuit of the regeneration amplifier whose duration is negligible-

20Ö8U / 1287

bar is small. Although the "1" individual pulse s (t) the amplitude threshold has already been exceeded before time t, but the "discovery" takes place (cf. the vertical Arrows) of the "1" individual pulse s (t) as a "1" representation only at time t, because it is due to the individual pulse of the time window signal emitted at time t IX the regenerator 16 connected downstream of the decision maker is opened »At this point in time, the regenerator detects 16 that a "1" single pulse is present. From this circumstance it follows imperatively if the duration T

of the square-wave single pulses of the digital signal (see. Fig. 5a) and the low-pass characteristics of the transmission unit are given that at time t = t + T the "1" single pulse s (t) is the predeterminable or has a predictable value i_ = s (t).

During actual operation of a regenerative booster, however, not only an isolated "1" single pulse s (t) occurs, but a disordered sequence of "1" or " ^w O" single pulses of the digital signal (such as, for example, in Fig. 5c) ο Because of the broadening of the individual pulses from T to more than 2T (see _o Fig. 5 *> and 5a) ψ due to the low-pass characteristic of the transmission unit, as mentioned, the individual pulses overlap as e.g. in Fig. 5d ( corresponding to the digital signal of Fig. 5c).

So I ₂ can be addressed as the predictable part of the overlap. Similarly, the amplitude s (t) = s (t -T) is referred to as the unpredictable part of the overlap in relation to the point in time t. In the case of the single pulse according to FIG. 5b, i applies because of its symmetry. = I ₂ *

2098U / 1287

_{In a manner known per se, the amount i 2 is} subtracted from the signal of the preliminary stage 122 of the araplitude regeneration circuit 12 from the time t to the time t ₁ (cf. FIG. 5e, where s (t) is denoted _{by i 2 in FIG} only added again when an ¹¹ O "single pulse is detected. This is the purpose of the overlap compensation device 125 of the amplitude regeneration circuit 12 in the regenerative amplifier of FIG. 2a by generating a corresponding compensation signal. by increasing the distance of the signal from the decision threshold.

This known measure therefore results in an increase in the storage spacing solely with regard to the foreseeable Proportion of the overlap reached.

2. Except for the near-end crosstalk interference just mentioned so-called extraneous disturbances occur, which can arise in different ways. For example, they show up in the vicinity of the terminals as originating from institutions with which the respective terminal must work together. If z. B. a PCM end parts is present, which generates the digital signal from a message signal, they can together with a telephone or telex exchange work · Sources of interference are then the switching equipment, i.e. relays and selectors. Furthermore can External interference on the transmission line in the form of an influence from a passing tram, Railway or power lines occur «Finally unbalanced dialing pulses can be sent to neighboring

2098U / 1287

2043UA

Coupling transmission lines as external interference. the External interference can therefore lie in a very broad frequency band, which makes it very difficult to suppress them.

The dimensioning of the equalizer of the individual regenerative amplifiers is also made considerably more difficult by the fact that the transmission link sections are usually predetermined and not of the same length. This is e.g. B. in telephone transmission lines in the form of cable lines due to the location of the cable ducts, in which w the regenerative boosters must be installed if existing, laid cables are to be used for the transmission of digital signals ₀ Since the position of the cable ducts is adapted to the local conditions, result there are inevitably differences of a few 100 m for a transmission link section with a total length of the individual transmission link sections of approx. 2 km to be bridged.

If the digital signal to be transmitted has components of very high frequencies at a very high pulse repetition frequency (up to a few MHz), the attenuation by the cable ^ can no longer be regarded as constant for all frequencies of the digital signal, which results in a distortion of the attenuation that increases with the length of the cable Digital signal through the cable expresses · If a longer transmission link section has to be bridged, it is not sufficient to increase the basic gain of the equalizer 11, but the gain of the high frequency components must also be increased. However, this also promotes the interference from near-end crosstalk coupling (see above) .

2098U / 1287

In Figo 6, the attenuation a of a local cable with 0.6 mm wire diameter, namely as a curve oC for 1 km, β for 2 km and ^ for 3 km section length, plotted over the frequency components of the digital signal

With regard to the different attenuation curves depending on the length of the transmission line sections, it is already known to provide different equalizers for the individual line sections. B ₀ eleven different plug-in equalizers must be in stock. However, this is not only very expensive, it can also easily be confused by telecommunications fitters who are not very familiar with electronics, so that the equalizer and transmission link section are assigned incorrectly.

So far it has mainly focused on the transmission of binary code encoded digital signals, but - as explained above - the difficulty of DC recovery Has.

This disadvantage of the binary code has led to the binary coded code words at the output in some cases of the transmitter-side encoder before being fed into the transmission path in code words of a higher-order code implement in which the code words consist of three or four-valued code symbols, while with binary code the code symbols are only two-valued. The code symbols or individual pulses of the digital signal can in this case thus assume three or four different states, o that one speaks of a ternary or quaternary code.

2098U / 1287

20431U

By using a ternary or quaternary code instead of the binary code, the desired freedom from direct current of the digital signal can be achieved if the redundancy of these codes compared to the redundancy-free binary code and only allows such sequences of code symbols, whose direct current components cancel each other out on average

are constant.

For example, ₀ B _{0 is} already a PCM transmission link

^ known (cf. ₀ z. B ₀ M, R, Aaron, PCM-Transmission in the Exchange Plant, Bell System Techn »Journal, January 19-62, pp. 125f 126) j in which the digital signal of the transmission path is encoded in the so-called bipolar code is. The bipolar code (see Fig. 1b) is a ternary code that arises from the binary code of the encoder as follows:

By comparing FIGS. 1a and 1b it can be seen immediately that the binary-coded digital signal of FIG. 1a merges into the digital signal -of FIG every second "1" single pulse with the voltage U is converted into a pulse with the voltage -U "The mean DC voltage content of the digital signal encoded in the bipolar code is zero, so that the zero line fluctuation that occurs in the binary code is avoided",

In the case of the bipolar code, the zero line migration is avoided, but it is caused by the loss of the low-frequency ones Portions of the digital signal encoded in the bipolar code in the transmitters Overshoots in the area the zero line, which degrade the signal-to-noise ratio.

2098U / 1287

In order to avoid this disadvantage, the lower limit frequency of the transmission unit of the regenerative power amplifier this known PCM transmission link are placed very low, which, however, has the consequence that Relatively low-frequency external interference can superimpose on the digital signal during transmission, which significantly reduced the signal to noise ratio of the digital signal from such interference

However, there are more potential changes in the bipolar code contained in the digital signal, so that the maximum of the power spectrum is already at half the single pulse repetition frequency lies, whereby the differentiator 131 in the control device 13 is omitted and more trigger pulses for the time regeneration circuit occur. Must have these benefits but with a much lower level of interference from extraneous interference - as already mentioned - and near-end crosstalk interference are paid · In addition, there is a little more effort in the amplitude regeneration and in the transmission stage necessary because two decision makers and two regenerators are necessary.

It is therefore the object of the invention, in a digital transmission link of the type mentioned above Regenerative boosters automatically have such a transfer function to give that the signal to noise ratio of the encoded in binary or any ternary or quaternary code Digital signal is increased

This problem is solved by the fact that the equalizer at least one actuator which determines its amplitude response has that the actuators at the output one with

2098U / 1287

204314-

- 2k -

connected to the output of the equalizer connected to the first control device, which the amplitude of a characteristic frequency component of the output signal of the equalizer and controls the actuators so, that the amplitude response of the equalizer of the respective Regeneratiwerstärkers the attenuation distortion of the previous transmission link section compensated »

^ The invention thus enables a stepless

^ Fit the regenerative booster to the length of the in

Transmission direction of the preceding transmission link section, so that a perfect equalization and thus a high signal-to-noise ratio is guaranteed in every case is. This advantage is also achieved without the need for an assortment of regenerative boosters, as was previously the case which leads to the difficulties mentioned at the beginning.

For an effective response of the first control device, it is expedient that the characteristic Frequency component, with the exception of the frequency zero, is the frequency of the digital signal at which the power spectrum of the digital signal has a maximum.

For circuit reasons it is advisable to that the equalizer consists of several equalizer filters connected in series with a fixed lower corner frequency, from which a high-pass characteristic begins, whose upper corner frequency does not affect the high-pass behavior can continue to increase, can be varied by the actuators.

2098U / 1287

20431

With a shift of the upper corner frequency, e.g. B ₀ upwards, the basic gain can be increased with »

As has already been explained, increases with increasing transmission link length because of the decreasing Reception level increases the near-end crosstalk, with the high-frequency Most disrupt the spectral components of the digital signal coming from the respective neighboring line »

An advantageous development of the Hype ^ agungss treks invention, in which the output of the regenerator each Regeneratiwers tärkers a Sendetiefpaß is filter-provided, therefore, is UAB the n-th transmission unit (n integer) consisting th n-from the Sendetiefpaßfilter of the regenerative amplifier, the (n + i) -th transmission link section and the equalizer of the (n + i) -th regenerative amplifier, has a low-pass characteristic, the upper limit frequency of which with decreasing amplitude of the characteristic frequency component of the input signal of the equalizer, d, h. as the length of the (n + i) -th transmission link section increases, through which the control elements of the captor can be displaced downwards.

Line adequate compensation of the attenuation distortion is achieved in most cases by the fact that the equalizer has only two equalizer filters, each of which contains an operational amplifier, each of which is preceded by a variable resistor as an actuator _t that the first equalizer filter has a high-pass behavior with constant amplitude response from a variable upper one Corner

2098U / 1287

frequency shows, at the same time the basic gain is displaceable, and that the second equalizer filter High pass behavior with a similarly changeable upper Has corner frequency and a fixed cutoff frequency from which the amplitude response drops.

It is also advantageous that the control resistors each to the inverting input of the operational amplifier are connected upstream.

The terminal of the variable resistor connected to the inverting input is then at an almost constant potential (/ virtual earth), which enables the control of the tellwiders tarids with an actuating signal that is dependent on digital signals; In particular, several variable resistors arranged and constructed in the same way can therefore be controlled with the same control signal ₍

It is also desirable that the variable resistors are field effect transistors whose control electrode is connected to the first control means ₀ field effect transistors have a better namely linear charac- ψ teristik than z. B, diodes and bipolar transistors and can be controlled without power.

If in an advantageous development of the invention the transmission units are designed as low-pass filters, whose (upper) cut-off frequency is decreased with increasing length of the transmission path, however the characteristic frequency component used for setting the equalizer is also more attenuated, what the first control device into which this fre-

2098U / 1287

20A3HA

frequency component is fed in as an actual signal, to be taken into account by ensuring that the amplitude is not kept constant, but rather during the setting of the equalizer on longer transmission link lengths decreases by the desired amount.

This desired amount can easily be adjusted by increasing the gain of a proportional control performing first control device is chosen so that for the amplitude of the characteristic frequency component of the output signal of the equalizer permanent control deviation occurs.

A particularly simple design of the first control device is that the first control device has an operational amplifier, of which the inverting input has an resistance to set the setpoint value grounded and the non-inverting input to a reference voltage source connected.

If the digital signal is encoded in binary code, its power spectrum is only maximal for the frequency zero. In order to still obtain a characteristic frequency component different from zero, the digital signal must be recoded by placing between the output of the equalizer and an actuating signal circuit of the first control device a differentiator is connected so that the characteristic frequency component is half the single pulse train or Clock frequency of the differentiated digital signal is,

When the coupling of each transmission line section to the two associated regenerators

209814/1287

20431U

is carried out by a transformer, it is recommended that that between the equalizer and the amplitude regeneration circuit there is a high-pass filter whose time constant is as small as possible compared to that of the transformer selected to be as large as possible is"

The installation of such a high pass has various advantages?

fe If on the transmission path, i. h "in larger

Distance from the terminals foreign interference, they have certainly low character, z _o B, influenced by a vorbeiführende tram, railway or power line. External interference can also appear as asymmetrical dialing pulses from neighboring transmission links, but because of the distance covered, they are already heavily worn out, so that they hardly contain any more spectral components at high frequencies. The high-pass filter provided according to the invention therefore eliminates this extraneous interference, but the low-frequency frequency components of the digital signal that are also lost as a result must be added to it again

ψ will be. Since the high-pass filter has a lower limit frequency due to its small time constants, which is significantly higher than that of the other necessary transmission elements with high-pass characteristics, especially the transformer, its influence can be neglected, so that when adding the lost digital signal frequency components only the time constant of the high-pass filter has to be taken into account is, which is very advantageous because the time constant of the high-pass filter, in contrast to the time constant of the transformer (cf. ₀ above, FIg ₀ 2b), is precisely defined »Da-

2098U / 1287

this improves the safety of the regenerative amplifier against low-frequency interference.

The high-pass filter also receives a direct-current separation in the equalizer from the Amplitudenregenerierkreis before, so that it is prevented that as a result of DC potential variations and the operating point of the input stage of the Amplitudenregenerierkreises varies ₀ If an additional galvanic separation can be carried out already in or before the equalizer, this would have the disadvantage the introduction of another, hard-to-define time constant in the regeneration booster »

When the digital signal is coded in binary code, must - as already explained - held a DC recovery "This is the regenerated digital signal at the output of the regenerator through a filter to the DC recovery, arranged in front of a decision maker precursor, both _s are" fed to Amplitudenregenerierkreis This direct current recovery is correct if every single pulse appears correctly recognized at the output of the regenerator. However, if an error occurs as a result of a disturbance, this error causes an incorrect compensation signal from the filter for direct current recovery in the input of the preliminary stage, which worsens the detection probability for the following individual pulses of the digital signal. During the time in which such an incorrect compensation signal occurs, a reduced immunity to interference of the regenerative amplifier must therefore be expected. On the other hand, since this time is determined by the time constant of the filter for direct current recovery and this in turn by the time con-

2098U / 1787

2043H4

constant of the high pass is determined, the time constant of the high pass should also be so small in this regard be chosen as possible.

If the transmission link according to the invention is provided is associated with a regenerator belonging to the amplitude regeneration circuit of each regenerative amplifier Overlap compensation device leading back to the input of the amplitude regeneration circuit for compensation

fc of the foreseeable portion of the overlap of individual pulses of the output signal of the equalizer that occurs on the trailing edge of the individual pulses, caused by a low-pass behavior of the relevant transmission unit, the n-th transmission unit (n integer) from the transmission low-pass filter of the η-th regenerative amplifier _t dem (n + i) -th transmission path section and the equalizer of the (n + i) -th regenerative amplifier, it is advantageous that the overlap compensation device has an overlap actuator, which is also connected to the first control device and is controllable so that the overlap compensation device only when it drops the amplitude of the characteristic

ψ Frequency component of the input signal of the equalizer,

d. h "with increasing length of the (n + i) -th transmission link section, can be actuated to a certain value is.

In the case of short lengths of the transmission link sections, interference from near-end crosstalk is excluded. The transmission unit is then adjustable through the The equalizer is optimally set if the (upper) cut-off frequency of the low-pass characteristic of the transmission

2098M / 1287

20431U

unit is so high that there is no overlap between adjacent individual pulses of the digital signal. Hence needs also no compensation for an overlap has to be made.

A simple embodiment of the overlap actuator is that the overlap actuator is a variable resistor is in the form of a field effect transistor, the control electrode of which is connected to the first control device is.

Another advantageous development of the invention is carried out by a delay distortion device for such a device that delays the low frequencies more than the high frequencies asymmetrical shaping of the individual pulses of the output signal of the equalizer that the foreseeable portion of the through a low-pass behavior of the relevant transmission unit caused overlap in favor of the unforeseeable, occurring on the leading edge of the individual pulses Proportion is increased.

Since the foreseeable portion of the overlap is compensated anyway, it does not matter if it is increased at the expense of the unpredictable portion _o The reduction of the unpredictable portion of the overlap that cannot be compensated also contributes to an increase in the signal-to-noise ratio .

In some cases it is even sufficient to use the delay time distorter by the properties of the previous transmission link section like a cable so that no special delay time distorter is required 1st.

2098U / 1287

20A31AA

The transmission path in which the time regeneration circuit each regenerative amplifier from a clock generation device tuned to the clock or single pulse repetition frequency and a downstream pulse shaper consists, and wherein the output of the equalizer is connected to the input of the clock recovery device via a non-linear filter which is derived from the output signal of the equalizer generates trigger pulses for the clock recovery device, is improved by the fact that the output the clock generation device via a second Regel- ™ device is fed back to the non-linear filter

In particular, it is expedient that the starting point of the non-linearity of the non-linear filter and thus the width of the trigger pulses can be controlled so that the energy of the trigger pulses just enough, the output signal of the clock recovery device to keep essentially constant »

The clock recovery device does not need because of their so loose as possible designed coupling to the digital signal So with excess energy, ie rela- length tively broad stimulus pulses which are derived from the digital signal to be controlled so that the clock recovery means fewer phase errors from the digital signal takes over. Since the phase errors are caused by interference or attenuation distortion of the digital signal, the measures just mentioned increase the signal-to-noise ratio with regard to the time regeneration of the digital signal. In addition, phase errors that could result from possible overdriving of the LC and crystal filters are avoided.

2098U / 1287

The non-linear filter is preferably a digital inverter, the threshold of which is determined by the second control device is movable.

a measured value uniform connects the output of the equalizer to the input of the non-linear filter it is advantageous that the second control device between the output of the clock generation device and a Input of the transducer has switched bias circuit through which the operating point of a to the transducer belonging operational amplifier is adjustable.

If the digital signal is encoded in binary code, the transducer can be used in a manner known per se be the second differentiator, which is also connected to the first Differentiator between the output of the equalizer and the control signal circuit of the first control device is connected, can coincide.

If the digital signal is encoded in bipolar code, the transducer can be a level shifter »

If the clock recovery device is based on the clock or single pulse repetition frequency of the digital signal has matched quartz filter, it is expedient that an LC filter, which is matched to the clock or single pulse repetition frequency of the digital signal, preceded with a low quality is.

Since the LC-FiIter has a low quality and therefore also has a low phase steepness, it sifts from the from

2096U / 1287

Nonlinear filter emitted impulses _s which can contain large phase errors, the clock frequency and thus removes from the impulses the operation of the quartz filter in the manner described above interfering frequency components. The LC filter generates a sinusoidal oscillation from each impulse, even if it is strongly damped.

It is also advantageous that the

ψ tive amplifiers provided digital operational amplifiers

Linking members are those of the well-known type Circuits with operational amplifiers have corresponding negative feedback networks.

In this way, extremely small dimensions of the regenerative boosters can be achieved, see above that the regenerative amplifier in a transmission path formed by an already laid cable in a Cable splice point can be accommodated within the sleeve (at the cable splice points are the commercially available Lengths of the individual cable lines connected to one another).

Another accommodation option for the regeneration boosters arises from the following considerations

Cables intended for the transmission of low-frequency signals are known to be provided with Pupin coils, which are arranged at intervals of about 2 km to make the frequency response in the low frequency range more even to design. Due to the Pupin coils, however, the frequency response is in the higher frequency range of the digital signals

2096U / 1287

very much attenuated so that the pupin coils removed anyway must be when you have very fast digital signals with a clock or single pulse repetition frequency of some MHz wants to transmit. For this reason, the regeneration boosters can already be used for the Pupin coils Existing housing can be used, so that the conversion of existing cables to the Transmission of digital signals is made much easier.

It is then also expedient that the digital links are integrated and have low power consumption corresponding to a low-power TTL series »

In this way you can easily use the regenerative boosters that are generally available low supply power.

After all, with a transmission link, where for each regeneration amplifier the amplitude regeneration circuit a decision maker showing a threshold value and the pulse shaper of the time regeneration circuit has a Schmitt trigger, it is advantageous that the decision maker, the non-linear filter and the Schmitt trigger are off digital logic gates are constructed, and that their trigger threshold serves as a reference voltage.

Such a design contributes to a further reduction in the spatial dimensions and reliability of the regeneration booster.

The invention is explained in more detail with reference to the drawing. Show it:

2098U / 1287

2043H4

1a and 1b a digital signal in binary or bipolar (Pseudo ternary) code;

1c shows the basic circuit diagram of known transmission links with regenerative amplifiers ;

2a shows a section of one of the transmission links from FIG. 1, the regenerative

m amplifier an essentially known one

Have structure;

2b shows an equivalent circuit diagram of part of the regenerative amplifier of Fig. 2a;

2c shows a non-linear filter according to Van der Houwen for obtaining the trigger pulses for the time regeneration circuit ;

FIG. 2d the signals belonging to the filter from FIG. 2c; FIG.

^ Fig. 3 at different points of the regeneration

signals appearing more strongly from FIG. 2a (except for signals VI and X);

FIG. 11 a, a sinusoidally approximated, binary-coded digital signal at the output of the equalizer; FIG.

Fig. Kb that this (Fig. 4a) j corresponding binarcodierte digital signal, and

Figo kc the power density of the digital signal of

2098U / 1287

2043H4

Pig. ka and the differentiated digital signal versus frequency;

Fig. 5a a single square pulse of the undistorted and undamped digital signal;

5b shows the actual course of a single pulse of the digital signal, due to the low-pass effect of the transmission unit;

Fig. 5c one of a sequence of several rectangular Ideal digital signal consisting of single pulses;

5d shows the actual course of the ideal digital signal of Fig. 5c, due to the low-pass effect the transmission unit;

Fig. 5e that partially compensated for overlaps Digital signal of Figure 5d;

5f shows the change in symmetry caused by delay time distortion a single pulse of the digital signal;

Fig. 5g and 5h, the phase amount ₀ or the running time of a real and ideal transmission cable as a function of frequency;

6 shows the attenuation as a function of the frequency for transmission link sections of different lengths from a local cable with a wire diameter of 0.6 mm and the amplitude response

2098U / 1287

20431U

of these transmission link sections associated equalizers of the invention regenerative enhancer used;

7 shows a block diagram of an exemplary embodiment of the regenerative booster used according to the invention for digital signals encoded in binary code;

™ Fig. 8a an embodiment of the equalizer of the

Regenerative booster of Fig. 7 »

FIG. 8b shows the basic circuit diagram of one of two filters forming the equalizer of FIG. 8a; FIG.

8c and 8d show the amplitude response of the first or second equalizer filter;

9a and 9b a so-called eye diagram at the transmitter output or after the transmission unit;

fc FIG. 10a shows an exemplary embodiment of the control device unit of the regenerative power boiler from FIG. 7 5

FIG. 10b is a basic control circuit diagram of the control device from FIG. 10a; FIG.

Figure _o 10c and 1Od the voltage signal amplitude of the digital signal component at half the clock frequency depending on the Übertragungsstreckenabachnittslänge for large or small gain of the control device;

208314/1287

" ³⁹ " 20431U

11a to 11d show the course of signals at different Points of the control device unit of Fig. 10a;

12 shows an embodiment of the time regeneration circuit, do h. the clock recovery device and the pulse generator, the regenerative amplifier of Fig. 75

13 shows an embodiment of the amplitude regeneration circuit of the regeneration booster from

Figure 7J

Fig. 1 ^ i an embodiment of the invention regenerative amplifier used for digital signals encoded in the bipolar code} and

15 shows the course of signals at various

Points of the embodiment of Fig.

In Fig. 71, an embodiment of the for the The regenerative amplifier used in the transmission path according to the invention for digital signals encoded in binary code shows, stages that correspond to the regenerative amplifier of FIG. 2a are given the same reference numerals provided, while deviations are followed by an "a".

A preferred embodiment of an in contrast to the known prior art according to the invention adjustable equalizer 11a basically consists of two equalizer connected in series according to FIG.

2098U / 1287

2CK3H4

filters 111 and 112, which have the structure shown in FIG. 8, the filter 111 having resistors R. and R ", a variable or variable resistor R _n , an operational amplifier V and a capacitor C, while the filter 112 has resistors R" to R_, a variable or regulating resistor R ₀ , capacitors C "and C" and an operational amplifier V_. (The variable resistances R ~ of the two filters 111 and 112 can be different.)

- The filters 111 and 112 are so by suitable

^ wired operational amplifiers V or V formed. Therefore, one can assume the equivalent circuit diagram of FIG. 8b for the filters 111 and 112, where U _E denotes the complex input voltage and U. the complex output voltage.

Because

^V A * ^U E "iS Ζ,

with (complex) Z and Z "results for the transfer behavior the filter 111 or 112 in general:

*! a = Z ₂ A ₁ ; ^u a [ ^iuJ ] «F (ju>),

_21st []

^U E. ^ EU OJ)

where F (jW) is the frequency response.

If the resistance or capacitance values have the same reference number as the associated components, the frequency response is the filter 111 or 112t

2098U / 1287

V ^R 1 1 + .HO ^R 0 ^R 1

R ₀₊ R ₁ ⁰ I.

F (, j OJ) = V UJtU (R ₀₊ R ₃₊ R ₅ ) C _{2 (2)}

I ₊ JCOR ₄ C ₃ )

If one forms the amount of the frequency response for a purely imaginary CU, i.e. the behavior of the filter in the case of forced continuous oscillations by means of excitation by a purely imaginary frequency, one obtains the so-called amplitude response ₀

The amplitude response | f | the filter 111 and 112 can be seen in FIGS. 8c and 8d as a so-called Bode diagram, ie the frequency f as the abscissa and the amplitude amount | F | The ordinates are in each case plotted logarithmically, with an approximation also being made by straight lines which meet at the so-called corner frequencies.

From Fig. 8c and d, these corner frequencies are f, f
f, f "or" f _o ', f', f ', f "' and E _Q 'are recognizable.
The analytical expressions for this are generally:

^f O = 2ψ R ~ C— ^{(3 a)}

^f 1 = -PfT ^R O ^{+ R} 1 (3 b)

^R O ^R 1 ^C 1

2098U / 1287

20A3UA

^f o '- 2 ir'

2tr (R ₀₊ R ₅ ) C ₂ (λ

V =

with the second for the double indexed frequencies
Digit of the index is assigned to a specific value of the variable resistor R _n .

In Fig. 8c and d each three different amplitude responses are shown. According to the invention, the different amplitude responses emerge continuously from each other by the variable resistor R _Q by adjusting signals from
a control device unit 13a (cf. also FIG. 7) provided according to the invention instead of the control device 13 of FIG. B ₀ 1 and 3 km can be assigned, which is desirable on the basis of the considerations made at the beginning. The acquisition of the control signals by the control device unit 13a is
yet to be explained.

According to FIG. 8c, the filter 11I has a high-pass behavior that begins at the lower corner frequency f _{n which is} independent of the variable resistor R "and is therefore fixed, corresponding to FIG
Time constant of resistor R and capacitor C
(cf. equation 3 ^a ) »that upwards through a variable
Corner frequency f _{1 (} f "or f" according to the time con

Il \ tZ. I __}

constant of the resistor R _Q and the capacitor C _{1 is} determined

2098U / 1287

2043U4

is (cf. equation 3 h) ₀ At the same time, the adjustment of the variable resistor R ~ also changes the basic amplitude response from f = O to L through vertical displacement »

The amplitude response of FIG ₀ 8d for the filter 112 deviates significantly from the amplitude response for the filter 111 shown in Fig. 8c from 'remains constant and a low-pass from the fixed cutoff frequency f _2' as the fundamental amplitude response of f = O and f corresponding to the time constants according to the resistor Rr and the capacitor C "occurs (see. Equation h c) _o The lower corner frequency f 'of the high-pass behavior is determined by the time constant of the resistor R" and the capacitor C ₂ (see. Equation h a), while the variable upper corner frequency ^ ₁₁ * f ' _1? or, f .. ο the time constant of the series connection of the resistor R and the variable resistor R ₀ together with the capacitor C "is equivalent (vglo equation k b).

The overall amplitude response of the adjustable equalizer 11a is composed of the amplitude responses for filters 111 and 112, which is a multiplication of the two amplitude responses F (j CO) according to equations (i) and (2) means. However, this multiplication corresponds to the Bode diagram because of the logarithmic plot of the amplitude response F (j CO) an addition of the respective amplitude responses. The total amplitude response of the equalizer 11a in the Bode diagram is thus obtained by adding the amplitude responses from FIGS. 8c and d. This sum is shown schematically in 6 for the three cases considered here, corresponding to a length of the transmission link sections of 1, 2 and 3 km shown

The overall frequency response of the η-th, by definition

2098U / 1287

- hk -

from the η-th transmission low-pass filter, the (n + i) -th transmission link section and the (n + i) th equalizer existing transmission unit results in the Bode diagram (cf. FIG. 6) from the difference between the amplitude response of the (n + i) th equalizer and the attenuation curve of the (n + i) th Transmission ^ section if the n-th transmission low-pass filter disregarded i.'rd, because this only becomes effective from half the clock frequency (here 1 .lüiz) «,

fc If this difference is almost equal to zero over the entire transmission frequency range, i.e. attenuation and The amplitude response curve in Fig. 6 almost coincide, there is no overlap of neighboring individual pulses of the digital signal, because there is no low-pass behavior with a cut-off frequency, which is less than half the single pulse train or clock frequency, shown by the transmission unit will. This is essentially the case for the 1 km section of the route. For 2 km length occurs at the Frequency 1 MHz (here half the single pulse repetition frequency or clock) already shows a clear difference (see. Fig. 6). So there is some low-pass behavior and thus a slight overlap of neighboring pulses »For

ψ 3 km in length, this difference is already very large, the low-pass behavior and the corresponding overlap of neighboring individual pulses of the digital signal are about twice as large. The latter circumstance is therefore favorable in order to optimally attenuate the near-end crosstalk interference that increases with the length of the transmission link sections and that occurs at higher frequencies.

If the transmission lines are formed by double-core lines of cables, since the cable ducts, especially in the case of the German Federal Post Office

2098U / 1287

about 300 - 500 m apart, the desired The regenerative amplifiers are arranged in the desired manner (described above), d. h in shorter Distance from one another in the vicinity of the end points and at a greater distance from one another when the distance is greater from the terminals.

The variable resistors R _n in the adjustable equalizer 11a are expediently formed by field effect transistors, the control electrode of which is supplied with the control signal from the control device unit 13a as a control voltage.

As explained at the beginning, e ^ is already known, due to the low-pass characteristics of the transfer units fc Conditional overlapping of individual pulses of the digital signal to compensate with regard to their foreseeable share (cf. Fig. 5a-e with the associated description) <, one Compensation or subtraction of the unforeseeable portion of the overlap is not possible, however According to this known prior art, however, according to the invention, the signal-to-noise ratio of the digital signal can thereby be achieved be increased so that the foreseeable portion to be subtracted later anyway, the overlap at the expense of not foreseeable proportion is increased, d. H. the symmetry of the individual pulses is changed. The enlargement of the predictable portion of the overlap is ultimately harmless

This change in symmetry can be achieved in a simple manner Time-of-flight distortion can be made in which the frequency components delayed differently »Figo 5f shows the influence of the delay time distortion on a symmetrical

2098U / 1287

2CK31U

tric single pulse of unpredictable overlap i and predictable overlap i.

During the practical testing of the invention it has surprisingly been found that a delay time distortion in the sense of the invention already by each transmission link section belonging to a transmission unit takes place when the transmission link is formed by cables.

In the case of an ideal cable used as a transmission link section without delay time distortion, the phase measure β increases linearly with the frequency, while the phase measure β of a real cable with delay time distortion is strongly negative at very low frequencies and only approaches the ideal curve at high frequencies. These courses are shown in FIG. 5g. After the transit time T ″ can be calculated from the cable length V , from the phase measure β and from the frequency, FIG. 5g shows FIG. 5h, in which the transit time TT of a real cable and the transit time of an ideal cable as a function of the angular frequency U) up're wearing. It is seen that the running ψ time X * of the ideal cable over the entire frequency range is constant, but the term V of the real cable is much larger at very low frequencies and at higher frequencies This behavior of the real cable helps to bring about the symmetry shift of the individual pulses of the digital signal according to the invention.

An enlargement of the solely caused by the Übertragungsstrekkenabschnitt delay distortion may be achieved by in or immediately behind the unloading

2098U / 1287

2CK3K4

Zerrer 11a a delay distortion device 20 (see. Fig. 7), z, B, all-pass elements (see ₀ Steinbuch / Rupprecht, Nachrichtenentechnik, 1967, page 115 ff o) is switched, which, like the cable, the lower frequencies stronger than the high frequencies delayed. The delay time distorter 20 can, however, also be arranged in the transmission low-pass filter of the preceding regenerative amplifier.

To the overlap compensation device 125 to Compensation for the foreseeable overlap of which a (still to be explained) embodiment shown in Fig. 13 is shown correctly, the foreseeable must first Proportion of the overlap can be measured.

A first possibility of measurement is that a rectangular single pulse (see. Fig. 5a) of duration T is sent through the transmission unit, so that from the impulse response s (t) (see FIG. 5b) the predictable Proportion i “of the overlap, especially with one Oscilloscope, can be read

The second possibility of measurement makes use of the so-called eye pattern, which is why should be explained in advance. The eye diagram is formed by a so-called "worst-case" analysis, d. H. the worst case of interference immunity is recorded by forming the eye diagram. To this end if a random generator sends a statistically disordered sequence of individual pulses over a transmission unit in the downstream amplitude regeneration circuit 12a of the regenerative amplifier in question.

In Fig. 9, for the sake of clarity, a

2098U / 1287

2043H4

fan case of the formation of the eye diagram (for binary coded digital signal). Fig. 9a shows for the case that the words to be transmitted by the digital signal encoded in binary code consist of three bits, all possible bit combinations and therefore words. These bit combinations or sequences of individual pulses of the digital signal experience - as mentioned - a broadening due to the transmission unit due to its low-pass behavior, which can be seen in each case in Fig. 9b. The overlay of all these sequences of broadened impulses at the output of the transmission unit then results in an eye diagram, the Z. B. can be observed with an oscilloscope. The eye diagram got its name, because in the middle there is a signal-free space, the "eye".

It should be evident that in comparison to the eye diagram at the transmitter output (see Fig. 9a), where the Eye as a borderline case is the middle square, the eye opening of the eye diagram after the transfer unit becomes smaller and smaller with increasing broadening of the individual impulses.

If you take off the eye diagram at such a point on the regenerative amplifier, at the point on the digital signal the overlap compensation device 125 has already acted, the optimal setting of the Overlap compensation device by choosing the largest possible Determine eye opening. With this second possibility of measuring the foreseeable proportion of the overlap the measurement is therefore immediately converted into the setting of the overlap compensation device.

2098U / 1287

2043H4

It should now an embodiment of the controller unit will be explained 13a of the regenerative amplifier used in the invention of FIG _o. 7

According to FIG. 7, the control device unit 13a consists of four blocks, namely the differentiator 131 ι one this controlling bias circuit 132, the non-linear Filter 135 and a Stellsigrialkreis 13 ^, the both are controlled by the differentiator 131. The control signal circuit 134 in turn controls the two equalization filters 111 and 112 as well as the overlap compensation device 125 of the amplification regeneration circuit 12.

According to the exemplary embodiment of the control device unit 13a of FIG. 10, the differentiator 13 5 contains an operational amplifier V connected as a differentiator, which converts the digital signal III in binary code into a digital signal U ^ (see FIG. 11a) or IV (see FIG. produced 3), which corresponds to a different code and its power spectrum maximum at half the clock frequency is Einzelimpulsfolge- or (cf. _o also Fig. 4a, 4b and 4c associated He purification). At the input of the operational amplifier V. lies a capacitor ^, -. "Further, a Riickkopplungswiders tand R _1., Available _u

The output quiescent _potential U Q (see «Fig. 11) of the operational amplifier V is a resistor Ii. _ the Vorspatmungskreises 132 to a certain value, z. B. by P.V.

As explained above, it is advantageous to use the transmission unit to be interpreted as a low-pass filter, its

2098U / 1287

2CU31U

(Upper) limit frequency with increasing transmission link length is lowered to near-end crosstalk interference However, this reduces the amplitude for the time regeneration and, according to the invention, also for the Adjustment of the equalizer 11a used as the aid of the frequency component equal to half the individual pulse train - or clock frequency more attenuated, which is what the control device unit 13a, into which this amplitude is fed as an actual signal has to be taken into account by ensuring that ^ λ c; that the amplitude is not kept constant, but rather when setting the equalizer to longer transmission link lengths decreases by the desired amount. This desired amount can easily be set if a proportional control is used o Γ This desired, remaining fluctuation in the amplitude half the clock frequency with different equalizer settings depended on either shortest or longest transmission link length is caused by the remaining control deviation U, definitely. The remaining control deviation is inversely proportional to the gain of the control device, as will be shown in more detail o

ψ The one for setting the equalization filters 111 and 112

and for controlling the overlap compensation device 125 serving control signal is as follows in the control signal circuit 13 ^ of Fig. 10a generates:

As an actual value for the extraction of the control signal, the amplitude of the differentiated binary-coded digital signal U or used IV, which (see. Also FIG. 6 with explanation) amplifiers representative of the length of the transmission line sections between the individual Regeneratiwer- is "The signal U ₁ is through a capacitor

2093U / 1287

- ^5t - 2CU31U

C. ^a ** f given the input of an inverting operational amplifier V, the gain of which is set by resistors R_ and Rg. The resistors R _- . g and R ₁₇ are dimensioned so that the desired characteristic of the equalizer 11a is taken into account (see ₀ below).

The operating point of the inverting amplifier V "is determined by the resistance R ₁ O so that its output quiescent voltage U ₁ Of) ^o f a fixed potential of, for.

B. V lies. The inverting amplifier V ₁ "can therefore, only the positive pulses of the signal evaluating U ₁₁ from which it at its output negative pulses U" wins (see _o Fig. 11c) _ς via a capacitor C and two diodes D, and D ₂ Build up a negative voltage Ug (cf. FIG. 11d) or (in FIG. 7) IVa at a capacitor Cg. The voltage U ₁ / -, the value of which is thus determined by the resistor Rg, is the desired control signal that _{actuates the variable resistors R 0} in the two equalization filters 11 1 and 112 and the variable resistor R " _Q in the overlap compensation device 125, with expediently all Variable resistors are field effect transistors.

The basic control circuit diagram of the control unit 13a with regard to its interaction with the adjustable equalizer 11a is shown in FIG. 10b, from which the control functions of the individual assemblies can be easily removed.

For a control loop with a proportional control device applies in general (cf. Merz, Basic Course in Control Engineering, 196 ^, pp. A / 52 ff.) for a control period t> 00:

2098U / 1287

2CK31U

TT ^S ο _Z 5 * Έ.

d ~ 1 + V .V V s ρ ρ

with ζ = control deviation occurring before control

U = system deviation occurring after regulation ("permanent" control deviation)

V = gain of the controlled system

V = gain of the control device o

If the gain V of the control device and thus of the amplifier V were large, according to the above formula the control loop for it so that with each setting of the equalizer half the clock or single pulse repetition frequency at the output of the equalizer with the same amplitude

occurs because for V> oo U, > 0 becomes.

¹ ρ d

In Fig. 10c and 10d, the amplitude of half the clock frequency U is in each case as a function of the length of the transmission link shown. The solid curve describes the course at the equalizer input, the dashed curve at the equalizer output,

In Fig. 10c the case of a large gain V is shown, i.e. h ₀ , the amplitude of half the clock frequency is the same for each equalizer setting. However, this is undesirable because, according to the invention, the transmission unit, consisting of the transmission stage of the preceding regenerative amplifier, transmission link section and equalizer, should have a low-pass characteristic, the (upper) limit frequency of which should migrate downward as the transmission link section length increases in order to better suppress the near-end crosstalk. The amplitude of half the clock frequency

2098U / 1787

^"53 - 2043H4

Therefore, when setting the equalizer, a large section length (for example
Equalizer output appear.

Section length (for example 3 km) reduced by U am

This is achieved according to the invention in that the Control device with the amplifier V "a limited, has low gain V. This increases its proportional range and the remaining control deviation U to the desired value (Figo 1Od).

The voltage U ₁ at the output of the operational amplifier V. also enters a digital inverter V (as the non-linear filter 135 in Fig. 7). The inverter V has a certain threshold, which is shown in FLg, 11a. From Fig. 11a it can also be seen that only those components of the signal U which are below the threshold of the inverter V let the inverter V respond.

il 1 tw

sen, so that it emits a signal U, which is made up of narrow composed of rectangular individual pulses. These ICinze! Impulses are used to control or initiate the subsequent time regeneration circuit, which by the clock warning device i'fa and the pulse shaper 15 are formed is, and are therefore dl ο An Antriebirnpu L so. Furthermore the voltage U corresponds to the signal V in FIGS. 3 and 7 o

In addition to setting the operating point of the operational amplifier V by means of the resistor R ", there is a second possibility of stimulating djosen work, namely via resistance 1! ₍ , R "and a diode U ,. Dea bias circuit 1Ί2. The diode D is connected to the output of the quartz filter \ h2 of the clock recovery oil device 1 ^ a via a pre-binding line provided according to the invention. As soon as the output signal VII

209814/1287

20A3H4

the quartz filter 1 12 exceeds * a certain value, the diode D becomes conductive, so that builds up on a capacitor C _ (see. Fig. 10a) of the bias circuit 132, a voltage more negative than the threshold (see ₀ Fig. 11a) of the inverter V _ ist _o This makes the output quiescent potential U of the operational amplifier V more positive, so that the negative pulse components of the voltage U ₁ only touch the threshold of the downstream inverter V or not at all U of the inverter V ₀ becomes increasingly narrower and finally disappears, as can be easily seen from FIGS. 11a and 11b when U is shifted upwards, _o

According to the invention is in the clock recovery device l ^ a the crystal filter 1 ^ 2 on the single pulse train or clock frequency matched active LC filter i4i connected upstream. The LC filter 1 * 1-1 has a low quality and thus a low phases :; division, so that it is derived from the Inverter V _ emitted impulses V filters out the clock frequency and therefore the function of the crystal filter 1 ^ 2 in the manner described at the outset, interfering frequency components removed from the trigger pulses V, i.e. from each The shock impulse is sinusoidal, even if it is strongly damped Vibration VI generated.

The from the output of the crystal filter 1 ^ 2 to the diode D of the preload circuit 13 'according to the invention leading connection liquid thus includes that of the differentiator 131 via the non-irradiating filter 135 and the clock recovery device i4a with the I.C-Fi L ter 1Ή and the quartz filter 1 ^ 2 running path of action, so that a second control loop is formed. This second control loop is

209814/1287

20431U

The width of the trigger pulses V and thus the energy that is fed from the non-linear filter 135 of the control device unit 13 & into the clock recovery device 1 ^ a and the time regeneration circuit is regulated via the detection of the output signal of the clock recovery device 1A. In contrast to the known control described above (cf. van der Houwen, op. The remaining amplitude fluctuations of the signal VII are insignificant, since the Schmitt trigger 151 of the pulse shaper 15, which is connected downstream of the quartz filter ih2, obtains the square-wave signal VIII therefrom.

In the following, an embodiment of the time regeneration circuit, consisting of the clock recovery device ika and the pulse shaper 15 _t , will be described in more detail with reference to FIG. 12 and the associated signals depicted in FIG.

The impulses of the signal V, which already contain the component of the clock or single pulse repetition frequency, arrive from the digital inverter Vp in the active LC filter 1 ^ 1, which consists of an operational amplifier V, which is fed back via a resistor R and whose input resistance is a damped series resonant circuit , which has a resistor R., a capacitor C _{? 1} and an inductance L ... At the output of the operational amplifier V ₂₁ , damped oscillations VI occur with the natural frequency 1 / T, which corresponds to the clock frequency. (see Fig. 3). A comparison of the signals V and VI shows

209814/1287

20A3H4

that the damped oscillations VI are repeatedly excited by the impulses of the signal V.

The damped oscillations VI reach the input of a second operational amplifier V_ ₂ , which is fed back (via a resistor βρς) and which contains a quartz Q as an input resistor, which is operated in series resonance. The circuit of the operational amplifier V ₂₂ forms the crystal filter 1 * f2. At the output of the operational amplifier V ₂₂ , the voltage VII arises, which is an almost uniform sinusoidal voltage with the frequency l / T, that is to say the single pulse train or clock frequency.

The signal VII arrives on the one hand in the bias circuit 132, as already mentioned, and on the other hand in the Schmitt trigger 151 t, which in FIG. 12 is composed of two non-elements V and V. and resistors R ₂ g and Rp7 ^# ^ ® ^r Schmitt trigger 151 obtains the square-wave signal VIII from the signal VII (cf. FIG. 3).

The signal VIII arrives at the input of the differentiating element 152, which consists of an operational amplifier V '_, whose feedback _{resistor was R 2q} and an input _{capacitor C "2} ". A resistor R ₂ Q ensures that the output of the differentiating element 152 shows 0 V in the idle state. This has the advantage that the differentiating element 152 can only evaluate the falling edges of the signal VIII and can obtain the time window signal IX with positive individual pulses therefrom. The time window signal IX is fed into the regenerator 16 (cf. FIGS. 7 and 13).

Finally, FIG. 13 shows an embodiment

2098U / 1287

of the amplitude regeneration circuit 12a in conjunction with the regenerator 16 of the Regenerati used according to the invention who stronger s.

The equalizer output signal III (cf. FIG. 3) is fed into the input of the preliminary stage 122 via a high-pass filter 121, provided according to the invention, consisting of a capacitor C and a resistor R ^ _{1, which feeds an operational amplifier V} ^ ₁ has.

The negative feedback operational amplifier V is in terms of DC voltage as a non-inverting amplifier with the gain factor unity for its trigger threshold voltage, which is the same as that of the downstream decision maker 123 in the form of an operational amplifier V_ "is designed. If (theoretically) no signal at the operational amplifier V "is applied, the threshold voltage therefore appears at its output. This exit is therefore on the same Potential like the threshold of the downstream operational amplifier V "". When the op amp V and V "are identical amplifiers or inverters the output potential of the preamplifier 122 and the input, threshold of decision maker 123 the same temperature response, so that the operating point of the preliminary stage 122 is stable in accordance with the threshold of the decision maker 123 is held.

However, these above considerations are only correct on the assumption that the preliminary stage 122 is DC separated from the equalizer 11a of the relevant regenerative amplifier *. Otherwise the operating point of the preliminary stage would fluctuate due to DC potential disturbances.

208814/1287

20431U

The desired galvanic separation is achieved according to the invention through the high-pass filter 121, that is to say immediately before the preliminary stage 122.

If additional galvanic isolation were to be carried out in or before the equalizer 11a, then this would have been the case the disadvantage of introducing another difficult-to-define time constant in the regenerative booster » According to the invention, however, this difficulty is overcome by the high-pass filter 121 arranged after the equalizer 11a, since the equalizer 11a can be implemented in galvanic coupling.

In this way you only need the indispensable time constants of the output transformer in the direction of transmission previous regenerative amplifier and the input transformer of the regenerative amplifier concerned to be considered.

The high-pass filter 121 also permits the avoidance of 18 and 10 of the known Regeneratiwerstärkers occurring ψ difficulties · According to the invention during the fixing of the time constants of the transformers are namely time * constant, the transmitter 10, and selected 18 as large as possible, while the time constant of the high-pass filter 121 small against which this transformer is and is only taken into account in the direct current recovery. This procedure is correct, since the high-pass filter 121 in the regeneration amplifier can be dimensioned precisely, and increases the safety of the regeneration amplifier against low-frequency interference.

Another advantage of the high pass 121 is shown when you consider the applied direct current recovery

2098U / 1287

2043U4

seeks. As has already been explained, the regenerated digital signal at the output of the regenerator 16 is fed via the filter 124 with direct current recovery to the preliminary stage 122 arranged in front of the decision maker 123. This direct current recovery works properly when each individual pulse appears correctly recognized at the output of the regenerator 16. However, if an error occurs as a result of a disturbance, this error causes a false signal from the filter 1J24, which is fed to the input of the preliminary stage 122 and the detection probability for the following individual pulses deteriorated. During the time in which such a false signal is effective, must So with a reduced immunity to interference of the regenerator a ti who are expected to be stronger.However, this time is determined by the time constant of the filter 12k and this in turn by the time constant of the high-pass filter 121. It is therefore advantageous to choose the time constant of the high-pass filter 121 as small as possible, as contemplated by the invention.

A similar procedure is used in the Regenerati according to the invention, who is more strongly applied to the foreseeable overlap to compensate ip of the individual pulses of the digital signal. In this case, too, it must be expected that an incorrectly recognized pulse will result in an incorrect compensation signal brings with it, which reduces the detection probability for the following impulse and on in this way, together with external disturbances, can bring further errors with it. So should in the event that large disruptions must be expected, so z. Am near the end points, the overlap compensation will not be applied. In this case one becomes the link section keep it short «So how

209814/1287

ORIGINAL INSPECTED

20431U

already explained, the characteristics of the adjustable equalizer 11b regulated according to the invention in such a way that after bridging a short stretch of road there is no overlap adjacent pulses occurs. Compensation for this overlap is then also not necessary and is not performed.

The preliminary stage 122, the high-pass filter 121, the overlap compensation device 125 and the filter for equal

Fe current recovery 124 form a unit in the exemplary embodiment of the invention, the central part of which is the operational amplifier V. The operational amplifier V is - as mentioned - designed as a summation amplifier, with the fixed feedback resistor R " ₂ and a summation point S, at which three signals are present according to the invention via appropriate evaluation networks." The evaluation network for the equalized digital signal VII is the high-pass filter 121 The (inverted) regenerated digital signal XIII is the filter 151, which acts as a low-pass filter, for direct current recovery. If these two evaluation networks are implemented with the same time constant, then there is good DC recovery

ψ achieved. To compensate for the foreseeable overlap i "of the individual pulses of the digital signal, the regenerated digital signal XIII is routed to the summation point S via an evaluation network in the form of the overlap compensation device 125.

In the overlap compensation device 125 the overlap compensation signal from the regenerated signal XIII at the output of the regenerator 16 via a simple Filter consisting of a resistor R ^ o and a Variable or regulating resistor R. », derived. The interlocutor

2098U / 1787

2043U4

The status R " _Q is adjusted by the control signal in the form of the voltage U ₁ ^ from the control signal circuit 13 ^, the resistor R" _{Q being} expediently a field effect transistor, into whose control electrode the voltage Uw- is fed.

According to the invention, a compensation is therefore automatically made the overlap of neighboring individual pulses of the digital signal only triggered when the same voltage U £ with regard to its two variable resistors R " controlled equalizer 11a (see. Fig. 8a) together with the other stages of the transmission unit according to the Length of the relevant transmission link section acts as a low-pass filter that causes the individual pulses to overlap (see the detailed explanations above),

In FIG. 13, the regenerator 16, which can generally be a bistable multivibrator, is a JK flip-flop with the control inputs J and K, a clock input cp and outputs q and q. The regenerated digital signal XIII occurs at the output q of the regenerator 16 _o

The filter 151 for direct current recovery, which is regenerated from the inverted, is connected to the output q Digital signal via two resistors R "(. And

R _{o /:} as well as a capacitor C__ the DC return gain-3 «32

nungs-Kompeneationesignal XI wins. By the DC regeneration compensation signal XI those low frequency components are added to the digital signal, which he previously lost through the transformer and the high-pass filter 121.

A practically proven embodiment of the

2098U / 1287

Regeneratiwerstärkera used according to the invention, from FIG. 7a, with the exception of the transmitter in stage 17, is completely implemented with digital TTL circuits, which, however, for the most part perform linear functions. This is achieved according to the invention in that, for. B ₀ a digital inverter (1/6 hex inverter, e.g. ₀ 74LO4 from Texas Instruments) according to the type of known circuits (cf. z, Bo W. 5 $ chaufelberger, operational amplifier and its ^ circuit technology, Kontron company, Munich-Feldmoching ) is fed back with fc operational amplifiers, taking into account that only the (existing) inverting input can be used and the (non-existing) non-inverting input is connected to a voltage that corresponds to the digital trigger threshold (approx. 1.4 V). One must also take into account that these amplifiers are not "ideal" operational amplifiers, e.g. B, input and output resistance and the low open gain considered. But they are particularly suitable for use in regenerative amplifiers, where speed is important with minimal power consumption at the same time, especially if you use circuits with low power consumption (Low Power TTL circuits).

Such a regenerative amplifier has been successfully operated for symmetrical, paper-insulated lines of 0.6 mm 0 with binary code and a single pulse or clock frequency of 2.0 ^ 8 Mbit / sec.

In Fig.] K is a Ausführungsbeispiei Regeneratiwerstärkers of the present invention used for the Ver r-processing ready encoded digital signals from the Bipolarcode (cf. 1b. Figo) = This embodiment has great similarities with the exemplary embodiment of FIG.

20981 A / 1 287

which - as already said - represents an exemplary embodiment of the regenerative amplifier used according to the invention for digital signals encoded in binary code (cf. FIG. _{1 a).} Signals occurring at various points in the embodiment of FIG. 14 are shown in FIG. 15.

Circuit stages of the embodiment of Figo 1 ^ which are identical to stages of the embodiment of Fig. 7 have been given the same reference numerals as in Fig. ₀ (with an addition "a" or "b" for the different signs of the bipolar signal), so that in this respect, an explanation of these stages can be dispensed with. A dash next to the reference number indicates an (explained) deviation. Similarly, signals on points in the exemplary embodiment in FIG. 14 that are identical to the exemplary embodiment in FIG. 14 are provided with the same Roman reference number and an additional prime.

Since a digital signal encoded in the bipolar code is a ternary signal, i.e. theoretically three different states occur in it compared to only two different states of digital signals encoded in the binary code, there is essentially a doubling of the stages of the amplitude regeneration circuit 12a, as can be seen in FIG.

A comparison of Figure 14 with Figure _s 7 also shows that compared to FIG 7, the filter to the DC power recovery 12 * 1 and the high-pass filter are disappeared 121, since -. As mentioned above - are DC-free in Bipolarcode encoded Digitaleignale. However, the last two stages mentioned can be provided if increased security against low-frequency interference is to be achieved.

2098U / 1787

■ ^6k "2CU3U4

In addition, compared to FIG. 7, the transmitter and the transmission low-pass filter have been separated, since the regenerated bipolar digital signal is only combined in the output transformer I8 ¹ . The transmission low-pass filter 17c, which leads directly to the next transmission link section, is connected to the output transformer I8 ^1.

The control device unit 13a ¹ of FIG. 1 ^ * differs from the control device unit 13a according to FIG. 7 only in that a level shifter 131 'is provided instead of the differentiator 131. A differentiation for recoding the output signal of the equalizer 11a is not necessary since this output signal has its power spectrum maximum at half the clock or single pulse repetition frequency in the case of the digital signal encoded in the bipolar code.

In the same way as the differentiator 131 in FIG. 7, the level shifter 13I ¹ , together with the non-linear filter 135, changes the width of the trigger pulses for the time regeneration circuit. The control signal circuit 13 ^ is connected directly to the output of the ψ equalizer 11a in this case. This is possible because the level shifter 131 ¹ does not carry out any recoding anyway. The level shifter 131 · may be similar to the differentiator 131 of Fig. 10a be constructed, but instead of the capacitor C a ¹ ohm shear resistance is provided.

2098U / 1287

Claims (1)

" ⁶⁵ " ■ 20A3HA Claims 1. Digital transmission line for the transmission of Digital signals, with regenerative amplifiers arranged between the individual transmission link sections for the regeneration of the damping distortion of the transmission link section arranged immediately beforehand in the transmission direction and noisy digital signals, the regeneration boosters for the incoming digital signals have an equalizing equalizer whose output is connected to one of the individual pulses of the output signal the equalizer pulses accordingly when a positive or negative amplitude threshold is exceeded the amplitude regeneration circuit generating the individual pulses of the undisturbed digital signal and to one of the phase errors of the output signal of the equalizer Time regeneration circuit is connected, both circuits controlling a regenerator for outputting the regenerated digital signal, characterized in that that the equalizer ("Ma) has at least one of its amplitude response determining actuator has that the actuators at the output of a connected to the output of the equalizer first control device (131 »1 3 * H 13 ^ ') connected which are the amplitude of a characteristic frequency component of the output signal of the equalizer detected and controls the actuators so that the amplitude response of the equalizer of the respective regenerative amplifier compensates for the attenuation distortion of the previous transmission link section (Fig. 7J 1 * 0 · 2, transmission link according to claim 1, characterized in that 2098U / 1287 20431U indicates that the characteristic frequency component, with the exception of the frequency zero, is that frequency of the digital signal at which the power spectrum of the digital signal has a maximum. 3. Transmission link according to claim 1 or 2, characterized characterized in that the equalizer (i1a) consists of several series-connected equalizer filters (111, 112) with a fixed lower corner frequency from which a high-pass characteristic begins, the upper corner frequency of which does not increase the high-pass behavior can be varied by the actuators (Fig. 7). k. Transmission route according to one of the preceding claims, wherein a transmission low-pass filter is provided at the output of the regenerator of each regenerative amplifier, characterized in that the n-th transmission unit (n integer), consisting of the transmission low-pass filter (17) of the η-th regenerative amplifier, the (n + i) -th transmission path section and the equalizer (i1a) of the (n + i) -th regenerative amplifier, has a low-pass characteristic, the upper limit frequency of which with decreasing amplitude of the characteristic frequency component of the input signal of the equalizer, i.e. with increasing length of the (n + i) - th transmission link section, through the actuators of the equalizer can be moved downwards (Fig. 2a, 7) 0 5. Transmission link according to claim h _t, characterized in that the equalizer (i1a) has two equalizer filters (111, 112), each of which contains an operational amplifier (V, V ₂ ), each of which has a variable resistor (R _Q ) 2098U / 1287 2043U4 is connected upstream as an actuator that the first equalizer filter (111) shows a high-pass behavior with constant amplitude response from a variable upper corner frequency Cf ₁₁ »f ₁₂ i f ..«), with the basic gain being shiftable at the same time (Fig. 8c), and that second equalizer filter (112) has a high-pass behavior rait a similarly variable upper corner frequency (f ', ^ ₁₂ ' »^ o ') ^and a fixed cut-off frequency (f ^ ¹ ) (Fig. 8d), from which the amplitude response drops (Fig. 7 , 8a). 6. Transmission link according to claim 5, characterized in that the variable resistors (R ₀ ) are each connected upstream of the inverting input of the operational amplifier (V ₁ , V) _l (Fig. 8a) ₀ 7. Transmission link according to claims 5 or 6, characterized in that the variable resistors (Rq) are field effect transistors, the control electrode with the first control device (13 * 0 is connected (Fig. 7 » 8a). 8. Transmission link according to one of the preceding claims, characterized in that the gain of the a proportional control performing first control device (13 * 0 is chosen so that for the amplitude of the characteristic frequency component of the output signal of the equalizer (i1a) a permanent control deviation (U,) occurs (Fig. 10b, 1Od). U * » 9. Transmission path according to one of the preceding Claims, characterized in that the first control 2098U / 1287 20431U device (13 ² O has an operational amplifier (V _{1 Ί} ), of which the inverting input is grounded via a resistor (jR ") to set the target value and the non-inverting input is connected to a reference voltage source (Fig. 10a, 10b, 1Od). 10. Transmission path according to one of the preceding Claims, wherein the digital signal is encoded in binary code, characterized in that between the output of the equalizer (i1a) and a control signal circuit (13Ό of the first control device, a first differentiator (131) is connected, and that the characteristic Frequency component is half the single pulse train or clock frequency of the digital signal (Fig. 4c, 7). 11. The transmission route to one of the preceding Claims j wherein the coupling of each transmission link section is made to the two associated regenerative amplifiers by transformers, thereby characterized in that between the equalizer (i1a) and the amplitude regeneration circuit (12) High-pass filter (121) whose time constant is as small as possible against the largest possible selected the transformer (10, 18) is (Fig. 7). 12. Transmission link according to one of the preceding claims, with one of the amplitude regeneration circuit of each regenerative amplifier belonging to its regenerator to the input of the amplitude regeneration circuit returning overlap compensation device to compensate for the predictable, on the trailing edge of the 2098U / 1287 - 69 - 20431U Individual pulses occurring portions of the overlap of individual pulses of the output signal of the equalizer caused by a low-pass behavior of the relevant transmission unit, the n-th transmission unit (n integer) consisting of the transmission low-pass filter of the η-th regenerative amplifier, the (n + i) -th transmission path section and the equalizer of the (n + i) -th regenerative amplifier, characterized in that the overlap compensation device (125) has an overlap actuator which is also connected to the first control device (13 ^) and is controllable so that the overlap compensation device only when the Amplitude of the characteristic frequency component of the input signal of the equalizer ("Ma), i.e. _e with increasing length of the (n + i) -th transmission link section, can be actuated to a certain value (Fig. 7) c 13. Transmission path according to claim 12, characterized in that the overlap actuator is a variable resistor (R " _o ) in the form of a field effect transistor, the control electrode of which is connected to the first control device (13 * 0 (Fig. 7, 13). 1 ^. Transmission path after a previous one Claims, wherein the n-th transmission unit (n integer) from the transmission low-pass filter of the η-th regenerative amplifier, the (n + i) -th link section fc and the equalizer of the (n + i) -th regeneration amplifier exists, marked t by one that delays the low frequencies more than the high frequencies Time-of-flight distorter (20) for such an asymmetrical form 2098U / 1287 20A3U4 mation of the individual pulses of the output signal of the equalizer (i1a), that the foreseeable portion (i ₂ ) of the overlap caused by a low-pass behavior of the relevant transmission unit, which occurs on the trailing edge of the individual pulses, in favor of the unforeseeable portion (I ₁ ) is increased (Fig. 5f, 7), 15 · transmission link according to claim 1 ^, characterized denotes overall ψ that the Laufzeitverzerrer is at least partially formed by the properties of the previous transmission line section (Fig. 5g "5h). 16. Transmission path according to one of the preceding Claims, wherein the time regeneration circuit of each regenerative amplifier consists of a clock frequency or a single pulse repetition frequency matched clock recovery device and a downstream pulse shaper, and wherein the output of the equalizer is connected to the input of the clock recovery device via a non-linear filter which generates trigger L pulses for the clock recovery device from the output signal of the equalizer, characterized in that the output of the clock recovery device (i4a) is via a second control device (132, 131} 132 ¹ , 131 ') is fed back to the non-linear filter (135) (Fig. 7, 1 * 0. 17. Transmission link according to claim 16, characterized in that by the second control device (135) the starting point of the non-linearity of the non-linear filter (135) and thus the width of the trigger pulses (v) is controllable in such a way that the energy of the 2098U / 1287 2CM31A4 impulse is just enough, the output signal (VIl) of the Clock recovery device (i ^ a) essentially constant to hold (Fig. 3, 7). 18. Transmission link according to claim 7 »characterized in that the non-linear filter (135) is a digital inverter (V ₁₂ ), the threshold of which is displaceable by the second control device (132, 131) (Fig. 7, 10a). 19 «Transmission link according to claim 18, wherein a transducer connects the output of the equalizer to the input of the non-linear filter, characterized in that the second control device connected between the output of the clock recovery device (i ^ + a) and an input of the transducer (131) Bias circuit (132) through which the operating point of an operational amplifier (V ₁₁ ) belonging to the transducer (131) can be set (FIG. 10a). 20. Transmission link according to claim 19 * wherein the The digital signal is encoded in binary code, characterized in that the transducer is in on is known to be a second differentiator (131) (Fig. 7). 21. Transmission link according to claim 10 and 20, characterized in that the first and second differentiators (131) coincide (Fig. 7). 22. Transmission link according to claim 19 »where the Digital signal is encoded in the bipolar code, thereby g e - 2098U / 1287 20431U ke η η ζ ο ic Ii π etj that the transducer is a level shifter (1 Ί 1 ') (cracked lh). 23. tibortraguiigsstreeke according to one of claims 1 <> to 22, Λ \ ίπ1) (! Ϊ dio Taktgewi iirmn [; beri cli t tmg a on the clock or egg π? El iinpu 1 sfol <; ei "requenz; dos Di gi t al si gnals matched. "Quarzl'dJier aufwoist, thereby geke η η ζ edc Ii η o. 1, that in the clock gain / vsei nri chtuiiß (i ^ a) the liuarzfiHej · (lh?) oin ani the clock oder Einsoldinj) ul si Olf.el'requeir / des Di ^ i tal si ßiial s abfjesi immtes LC filter (i'n) mi i (; eriiif; er (Jiite- vorgpscha] tnt is (Fiß.?). trei -ho. juich eiiKini of claims h to, '? 3, dadurcli ß <; l <onn ^ ed chiiet, because <the operational amplifiers are dißiinln Vei'linkplnnßsßl i or that after the Λι-t of the received S dial dip with operational amplifiers Corresponding (Jef.cnkopplun ^ s networks aui sen. . '? 3 <Ül) ei'trnßiiii /; HRtrecke according to claim 2 ^ ₅ thereby ßekennzei without i , dal! the digital connection elements are inte- l'vi ei't and have a power consumption similar to that of a low-power TTL series. B ^>. iJbeiti'agungsstube nach Aiisi) ruch 2k or 2'5 | whereby for every Hegenerat i vvers Turk the Anipli tudon regeneration circuit has a syllable 1 expanding I / iitscheider and the pulse formula of the time generator circuit has a Schmitt trigger, characterized in that the decision! 1Ί5) "" id the Schmi tt-Tj'i gger (151) are built up from digital logic elements, and that their trigger threshold served as reference voltage 2098U / 1787 ORIGINAL INSPECTED