 |
Description  |
|
|
BACKGROUND OF THE INVENTION
The invention relates to a self-adjusting equalizer configuration which
automatically adjusts to the cable length. The equalizer configuration
includes a band limitation for the transmission of digital signals and
consists of an equalizing amplifier, a correction filter, and an amplitude
control loop which includes a comparison of actual and nominal values.
Digital signal regenerators in transmission systems with coaxial cables as
the transmission medium have an automatic attenuation equalizer at the
input in order to equalize the distortions resulting from the
frequency-dependent cable loss. The amplitude of the equalized signal;
e.g., the amplitude of the signal at the discriminator, is used as a
control criteria for adjusting the equalizer to the length of the cable.
This pulse amplitude must be as independent as possible from the input
pattern. However, this is the case only when a specific band limitation is
adhered to in the equalizer. To suppress random noise and interference
signals, however, a tighter band limitation must be used which results in
varying pulse amplitudes in the signal path as a function of the input
pattern.
The band limitation required to reduce noise is usually implemented as a
"roll-off" band limitation as described, for example, in the book by P.
Bocker: "Datenuebertragung" (Data Transmission), Vol. 1, pages 98-114
(Springer Verlag 1978).
Transmission systems for higher bit rates, such as 34 Mbit/s, are only
possible with a roll-off factor that is less than 1 for the field lengths
and transmitting power applicable in this case, so that r<1. This fact
results in signal overshoots, the amplitude of which is dependent on the
pulse sequence. This concept will be explained later with reference to
FIG. 1. Another known configuration that circumvents the disadvantage of
the dependence of the equalizer setting on the pulse sequence will be
explained with reference to FIG. 2. The difficulties that are particularly
evident for this configuration are explained as follows: The maximum gain
of the equalizing amplifier for a roll-off factor with the value 1 is
greater and takes place at a higher frequency than that of an amplifier
for a roll-off factor with a value <1. This is illustrated in FIG. 7 and
requires more effort, particularly at higher bit rates such as 140 Mbit/s.
In addition, two phase-equalized filters are located in the signal path,
the tolerances and aging of which directly influence the accuracy, and
therefore the quality, of the regenerator.
SUMMARY OF THE INVENTION
The object of the invention is to provide an automatic equalizer
configuration in which the correction filter is not located in the signal
path, but in the amplitude control circuit, so that fewer demands must be
made on the accuracy of the correction. Moreover, the equalizing amplifier
only requires a band limitation with a roll-off factor <1. This also
ensures that the correction filter attenuates instabilities of the entire
control circuit for frequencies above the transmission band due to its low
pass response.
In general, the invention features a self-adjusting equalizer configuration
which automatically adjusts to the cable length, with band limitation for
transmitting digital signals, and which includes an equalizing amplifier,
a correction filter and an amplitude control loop with comparison of
actual and nominal values, in which the equalizer amplifier is located in
the signal path, and the correction filter is located in the amplitude
control loop so that the output signal of the correction filter represents
the actual value for the amplitude control loop.
In preferred embodiments the eqalizer configuration includes a signal pulse
derived after the correction filter, to control a discriminator connected
in series with the equalizer amplifier; the circuit configuration and the
size of the correction filter are selected so that it has the following
transfer function
##EQU1##
where, V.sub.o is basic amplification, f.sub.1 is the cut-off frequency of
a low pass filter, and f.sub.R is a resident frequency of a tuned circuit,
with a and b as its constants.
Other features and advantages of the present invention will become apparent
from the following detailed description, and from the claims.
For a full understanding of the present invention, reference should now be
made to the following detailed description and to the accompanying
drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a block diagram of an existing regenerator described in the book
by Bocker entitled "Datenuebertragung" (Data Transmission) which has
already discussed.
FIG. 2 is a block diagram of the previously discussed equalizer, which has
a band limitation for a roll-off factor of r.apprxeq.1 and a correction
filter for r=1 or r=0.7 in the signal path.
FIG. 3 is a block diagram of the equalizer configuration of the invention
with a correction filter for r=0.7 or r=1 in the amplitude control loop.
FIG. 4 shows an eye diagram of the signal after the discriminator and after
the roll-off converter. FIG. 4a shows these relationships for a roll-off
factor r=0.7 and FIG. 4b shows these relationships for a roll-off factor
r=1.
FIG. 5 shows the theoretical effective noise band with B (r) and the band
width gain G(r)=10 Log B(r=1)/B(r), referenced to r=1 and as a function of
the roll-off factor r of an equalizing amplifier for a line with
##EQU2##
FIG. 6 shows the theoretical gain of the equalizer for r=1 and r=0.7.
FIG. 7 shows the transfer coefficient Ue of line and equalizer for a
roll-off factor r=1 and r=0.7 as a function of the frequency f.
FIG. 8 shows a theoretical correction filter K/dB for reducing the roll-off
factor as a function of the frequency; for different roll-off factors
r=0.5, 0.7, and 0.9.
FIG. 9 shows a theoretical correction filter K/dB for reducing the roll-off
factor as a function of the frequency f (the Nyquist frequency f.sub.o
=12.88 MHz).
FIG. 10 shows the measured gain V.sub.E /dB as a function of the frequency;
curve a shows the gain of the attenuation equalizer and curve b shows the
gain of the attenuation equalizer, including the roll-off converter.
FIG. 11 shows the measured transfer coefficient V/dB of the entire
transmission path. Curve a shows the transfer coefficient V of line and
attenuation equalizer and curve b shows the transfer coefficient of line,
attenuation equalizer and roll-off converter.
FIG. 12 shows the measured gain V/dB of the roll-off converter for a
roll-off factor of r.apprxeq.0.7, which is converted to a roll-off factor
of r.apprxeq.1.
FIGS. 13A-13D show schematically possibilities for implementing a roll-off
converter, and how it can be utilized for the automatic equalizer
described here. FIG. 13a shows a transistor as an active element; FIG. 13b
also shows a transistor as an active element, where FIG. 13b illustrates
the dual circuit for FIG. 13a. FIG. 13c illustrates an operational
amplifier as an active element in which V goes toward infinity; and FIG.
13d illustrates an additional circuit option with an operational
amplifier. The transfer response U.sub.2 /U.sub.1 is shown directly and
the transfer coefficient of the circuit according to FIG. 13c is equal to
that of FIG. 13a, while the transfer coefficient of the circuit according
to FIG. 13d is equal to that of FIG. 13b.
DETAILED DESCRIPTION
It should be noted that elements having the same function are assigned the
same reference numbers in the individual figures so that explanations
given for these elements will be valid for all figures.
Referring to FIG. 1, data signal 10 reaches equalizer 1 which has band
limitation. Output signal E or output signal E' appears at the output of
this equalizer. Output signal E is applied to discriminator 3 which is
followed by final amplifier 5. In amplitude control loop 2', the output
signal E is also applied to amplitude control loop 2 from which, in turn,
band-limited equalizer 1 is triggered. An additional signal E' triggers
timing recovery generator 4 which controls discriminator 3.
A similar configuration can also be seen in FIG. 2, where the input signal
is again identified by 10 and the output signal by E. Amplitude control
loop 2' can also be recognized along with signal E', which is connected to
amplitude control loop 2'. In contrast with FIG. 1, correction filter 6 is
now provided in the signal path in the circuit diagram of FIG. 2. Details
concerning the effect of this correction filter will be given later.
In the embodiment shown in FIG. 3, correction filter 6 is located in
amplitude control loop 2'-2. Therefore, output signal E is applied to ths
correction filter and subsequently reaches amplitude control circuit 2. A
roll-of factor of r=0.7 in the signal path 10/E is assumed and a roll-off
factor r.apprxeq.1 results at the output of correction filter 6 or at the
input of amplitude control circuit 2. It is also possible there, as in
FIG. 2, to provide an output signal E' after correction filter 6.
FIG. 13 illustrates the simple options which can be used to implement the
transfer functions required here. For this purpose, a low pass function
and a resonant amplifier with a series and parallel attenuation circuit
must be used. In the circuits shown in FIG. 13, the transfer function
U2/U1 is directly specified at the individual circuits. U2 is the output
voltage and U1 is the input voltage of an amplifier. In FIG. 13a, a
transistor T is used as the active element with input voltage U1 applied
to the base of the transistor. The following are located in the emitter
circuit: the parallel circuit consisting of a resistor R.sub.p, and a
series-circuit consisting of capacitor C, coil L, and dissipation resistor
R.sub.s. Resistor R.sub.c and capacitor C.sub.c are located in the
collector circuit. FIG. 13b illustrates a possible circuit which is the
dual circuit for the circuit shown in FIG. 13a. For this reason, the
series-circuit in the emitter circuit consists of resistor R.sub.c and
coil L.sub.E, while the series configuration consisting of a resistor
R.sub.s and the series-connected parallel circuit consisting of a coil L,
a resistor R.sub.p, and a capacitor C are located in the collector
circuit.
In FIGS. 13c and 13d, amplifiers are used which have the characteristics of
operational amplifiers and are therefore provided with the reference
symbol Op. Consideration should be given to the fact that the gain v goes
toward infinity. In the circuit illustrated in FIG. 13c, the parallel
circuit consisting of a resistor and a capacitor is located in the
negative feedback circuit, for example between the amplifier output and
the amplifier input. The parallel circuit consisting of a resistor and an
attenuated series resonance circuit is located in the amplifier input.
In the circuit illustrated by FIG. 13d, the series circuit consisting of a
resistor, to which the parallel circuit consisting of a capacitor, a
resistor, and a coil is connected in series, is in a negative feedback
circuit, while the series circuit consisting of a coil and a resistor is
located in the amplifier input.
Reference is now made to the following configurations for further
explanation and for a better theoretical understanding of the invention.
FIG. 1 shows the block diagram of a typical regenerator, consisting of line
attenuation equalizer 1 with amplitude control circuit 2, discriminator 3
with timing recovery 4, and final amplifier 5.
The band limitation required for reducing noise (receiver noise, cross-talk
interference) before the discriminator is usually implemented as a
roll-off band limitation according to a previously discussed reference.
Transmission systems for higher bit rates (such as 34 Mbit/s) for the
field lengths and transmitting power specified therein can only be
implemented with a roll-off factor of r<1.
However, a factor of r<1 results in signal overshoots, the amplitude of
which is dependent on the pulse sequence. This means that the peak value
rectifier in the amplitude control circuit of the equalizer supplies a
different value for random sequences than for a fixed pulse sequence, and
that it supplies different values for different sequences. Therefore, the
automatic equalizer (provided the time constants in the control circuits
are sufficiently long) will set itself to the correct value only in the
case of random sequences or in the case of a particular fixed sequence.
FIG. 2 illustrates a configuration known in principle which avoids the
disadvantage of setting the equalizer as a function of the pulse sequence.
Equalizer 1 contains a band limitation with r.apprxeq.1 and its output
signal (E) is virtually free of overshoots. The timing recovery for
reducing the ground jitter of the regenerator can also be derived from
this signal E' as needed.
In the signal path, equalizer 1 is followed by additional filter 6 with a
linear phase for reducing the roll-off factor (FIG. 6 shows the
theoretical value of the transfer function).
The disadvantages of the configurations shown in FIGS. 1 and 2 are as
follows:
(a) The maximum amplification of the equalizing amplifier for r=1 is
greater and takes place at a higher frequency than that of an amplifier
for r<1 (FIG. 7) and requires more effort, particularly at high bit rates
such as 140 Mbit/s.
(b) Two phase-equalized filters are located in the signal path, the
tolerances and aging of which directly influence the accuracy of
equalization (nodes in the diagram), and therefore the quality of the
regenerator.
The configuration shown in FIG. 3 avoids these disadvantages, in accordance
with the invention.
Correction filter 6 for roll-off conversion is not located in the signal
path, but in amplitude control circuit 2, 2', thus requiring less demands
to be made on the accuracy of the correction. Equalizing amplifier 1 is
implemented only for a band limitation with 1<r.
A 34 Mbit/s theoretical transmission system with a cable loss a.sub.o =60
dB at the Nyquist frequency f.sub.o =12.88 MHz will be used as an example.
The theoretical transfer function of line attentuation equalizer 1 is
##EQU3##
with the desired receive spectrun of a roll-off pulse with band
limitation,
##EQU4##
(A.sub.e is the receive amplitude, r is the roll-off factor) the spectrum
of a transmit pulse which approximates actual applications, (A.sub.s is
the transmit amplitude; the 2nd factor is the spectrum of an ideal,
half-bit-wide, square wave pulse; the 3rd factor takes into account
cosine-shaped signal slopes), and the approximate transfer function of a
line
##EQU5##
FIG. 6 shows the gain V=20.multidot.log F.sub.E (f) of the equalizer for
r=1 and r=0.7. FIG. 7 shows the transfer coefficient of the entire
transmission path u=20.multidot.log (F.sub.E .multidot.F.sub.Ltg).
R=0.7 has been chosen for the example.
The theoretical transfer function F.sub.K1 for the correction filter
illustrated by FIG. 2 for reducing the roll-off factor is
##EQU6##
FIG. 8 shows 20.multidot.log F.sub.K1 for different r values.
The theoretical transfer function for the correction filter illustrated by
FIG. 3 for increasing the roll-off factor is
##EQU7##
FIG. 5 shows 20.multidot.log F.sub.K for different r values.
Now an actual transmission system according to FIG. 3 will be examined.
This system has a cable loss of 60 dB at the Nyquist frequency f.sub.o
=12.88 MHz since the previously explained theoretical transfer functions
F.sub.E (f), F.sub.K1, and F.sub.K can, of course, only be approximated.
The line attenuation equalizers for this system are implemented as follows:
A workable compromise between the effort and accuracy of equalization is
the transfer function
##EQU8##
This function is implemented in the embodiment by two tuned amplification
stages with a serial and parallel attenuated resonant circuit (resonant
frequency f.sub.R, circuit constants a and b), a passive high pass filter
(cut-off frequencies f.sub.1 and f.sub.2), and a passive all-pass filter
(characteristic frequency f.sub.a) for the linearization of the phase of
the transmission path.
In FIG. 10, curve a shows the measured gain of the line attenuation
equalizer. In FIG. 11, curve a shows the measured gain of the series
connection consisting of line plus attenuation equalizer.
The roll-off factor is r.apprxeq.0.7.
FIG. 4a shows the associated diagram for a "pseudo" random pulse sequence
with its pattern-dependent overshoots.
This output signal is barely suitable as a control criterion for regulating
a variable attenuation equalizer.
An equalizer for converting the roll-off from rf 0.7 to r.apprxeq.1 is
implemented as follows.
The theoretical transfer coefficient shown in FIG. 10 need only be
approximated up to a gain of approximately 10 dB; as shown by the
measurements. The following transfer function is sufficient for this
purpose:
##EQU9##
Phase equalization is not required.
FIG. 12 shows the transfer coefficient measured using f.sub.1 =12.3 MHz,
f.sub.R =20 MHz, V.sub.o =1, a=1.17, and b=0.16.
FIG. 4b shows the diagram as measured after the converter. This signal is
suitable as a control criterion.
The previously mentioned transfer coefficient function is composed of a low
pass filter and a resonant amplifier with a series and parallel
attenuation circuit.
FIG. 13, which has already been described, shows simple possibilities for
implementing this function.
There has thus been shown and described a novel self-adjusting equalizer
configuration which automatically adjusts to the cable length which
fulfills all the object and advantages sought. Many changes,
modifications, variations and other uses and application of the subject
invention will, however, become apparent to those skilled in the art after
considering this specification which discloses embodiments thereof. All
such changes, modifications, variations and other uses and applications
which do not depart from the spirit and scope of the invention are deemed
to be covered by the invention which is limited only by the claims which
follow.
* * * * *
|
|
 |
|
 |
Description |
|
