JPH0611678Y2 - Automatic equalizer - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial field of application] The present invention transmits a signal waveform distortion due to a loss characteristic of a transmission line in order to restore a transmission signal without error on the receiving side in digital subscriber line transmission. The present invention relates to an automatic equalizer that automatically equalizes according to the status of a track.

[Prior Art] Conventionally, as a technology in such a field, for example, there is one described in the following documents.

Reference 1; Proc. 1993, IEICE Communications Division National Convention Proceedings 133 (Sho 59) “Digital Signal Processing Automatic Equalization LSI Configuration” Reference 2: Proc. 1985 IEICE General Conference Proceedings 2081 ( 60) "Study on equalization circuit in 320 kb / s ping-pong transmission system" Document 3; Japanese Patent Laid-Open No. 54-84952. In the past, this type of automatic equalizer was often composed of an analog circuit. Recently, with the digitalization of communication circuits in general, as described in the above-mentioned document 1, a digital system having an advantage that a large-scale integrated circuit (hereinafter, referred to as LSI) can be easily performed and processing can be flexibly performed. A configuration by processing is being considered.

FIG. 2 is a block diagram showing an example of the configuration of an automatic equalizer using a digital processing circuit that is generally used conventionally.

This automatic equalizer has an input terminal 1 and an output terminal 2,
A pre-filter 3 and an analog / digital converter (hereinafter referred to as an A / D converter) between its input and output terminals 1 and 2.
4 and √AGC5 and roll-off filter
ter) 6 and the digital processing circuit are connected in cascade. A control circuit 7 is provided on the output side of the roll-off filter 6.
Are connected, and the characteristic of the √AGC 5 is controlled by the control circuit 7.

The pre-filter 3 is a low-pass filter (hereinafter referred to as an LPF) formed of an analog circuit, and has a function of reducing aliasing (aliasing, frequency overlap) due to signal sampling. The A / D converter 4 is a circuit that samples and digitizes the output signal of the pre-filter 3 to convert it into a digital signal, and supplies it to the √AGC 5.

The √AGC 5 is a circuit that compensates the loss characteristic of the line based on the control signal output from the control circuit 7, and functions as a high-pass filter (hereinafter referred to as HPF). The roll-off filter 6 functions to attenuate frequency components outside the signal band while suppressing intersymbol interference. These √
The digital processing circuit including the AGC 5 and the roll-off filter 6 is configured by a digital signal processor (hereinafter referred to as DSP), a wired logic, or the like, as described in Document 1 above.

In the above configuration, when the received signal SI is input to the input terminal 1, the received signal SI is
The band is limited so that the frequency component of 1/2 or more of the sampling clock frequency in the / D converter 4 becomes negligibly small. The output signal of the pre-filter 3 is converted into a digital signal by the A / D converter 4, and then equalized by the √AGC 5 and the roll-off filter 6. As a result, the equalized output signal So with small distortion is obtained from the output terminal 2. The output signal So is a digital signal, but an analog signal can be obtained by performing digital / analog conversion (hereinafter referred to as D / A conversion) on the output signal So.

Waveform distortion due to line loss characteristics varies depending on the type and length of the line. Therefore, in the automatic equalizer of FIG. 2, the control circuit 7
Therefore, the characteristics of √AGC5 are set so that the best equalization can be performed according to the situation of the line. Usually, as this adaptive algorithm, since the received signal amplitude and the line loss characteristic are closely related, the characteristics of √AGC5 are obtained in advance for some reference lines and the output of the equalizer is calculated. A method is adopted in which the characteristic is selected so that the amplitude value of the signal So falls within the reference range.

By the way, in this kind of automatic equalizer, it is usually necessary to perform two functions of equalization of waveform distortion due to line loss characteristics and elimination of out-of-band noise. In the conventional automatic equalizer, there is a tendency for the equalizer as a whole to perform these operations without clearly distinguishing these actions. Specifically, the roll-off filter 6 has a small interference at the time of data identification of the output signal So, such as a raised cosine pulse signal, based on the assumption that the reverse characteristic of the line is realized by another circuit, and The characteristics are determined so that the signal has a small frequency band. And √AGC5 is the pre-filter 3
So that the overall characteristics of and are the reverse characteristics of the target line,
It defines the characteristics. That is, the action of is performed by the entire equalizer, and the action of is performed by a circuit other than √AGC5 which is the HPF.

[Problems to be Solved by the Invention] However, in the automatic equalizer configured as described above, √AGC5
Since both the roll-off filter 6 and the roll-off filter 6 need to perform highly accurate waveform approximation in the time domain, there is a problem that the circuit tends to be complicated and the circuit scale becomes large. Particularly, when a large removal effect is required for noise having a frequency component such that (clock frequency bandwidth of digital processing circuit) >> (signal bandwidth), an extremely complicated roll-off filter 6 is required, This tendency becomes remarkable. Further, there is a problem that design is difficult because mutual interference between the circuits is large.

In order to solve these problems, it is conceivable to configure the digital processing circuit including the √AGC 5 and the roll-off filter 6 with an analog circuit, as described in Document 3 above. However, if the digital processing circuit is configured by an analog circuit, not only the configuration of the control circuit 7 becomes complicated, but also the equalization accuracy changes depending on the characteristics of the circuit constituent elements, so that the design is difficult and still technically satisfactory. I couldn't get a better automatic equalizer.

The present invention provides an automatic equalizer that solves the problems of the above-mentioned prior art, that is, the increase in circuit scale and the difficulty in design.

[Means for Solving Problems] In order to solve the above problems, the present invention provides a pre-filter configured by an LPF of an analog circuit for band limiting an input signal, and an analog signal output from the pre-filter. And an A / D converter for converting the
A digital processing circuit for equalizing the output signal of the D converter;
In a digital subscriber line automatic equalizer comprising a control circuit for controlling the characteristics of the digital processing circuit based on the amplitude value of the output signal of the digital processing circuit, the pre-filter is provided in the digital processing circuit. The configuration is such that frequency components above the clock frequency band are attenuated.

Further, the digital processing circuit has a variable filter whose characteristic is variably controlled by the control circuit according to a line condition, and which equalizes the line loss characteristic in the signal band with respect to the output signal of the A / D converter. , A fixed phase filter which has a constant characteristic regardless of the line condition and which attenuates frequency components higher than the signal band and lower than the clock frequency band.

[Operation] According to the present invention, since the automatic equalizer is configured as described above, the input signal is attenuated by the pre-filter in the frequency component above the clock frequency band of the digital processing circuit to reduce aliasing. , A / D converter converts the signal into a digital signal, which is then sent to the variable filter. The characteristic of the variable filter is controlled by the control circuit, the waveform distortion is equalized by the line loss characteristic in the signal band, and the equalization result is sent to the fixed filter. The fixed filter removes out-of-band noise by attenuating frequency components higher than the signal band and lower than the clock frequency band of the digital processing circuit without adversely affecting the frequency components in the signal band. As a result, the circuit scale can be reduced and the design can be facilitated. Therefore, the above problems can be eliminated.

[Embodiment] FIG. 1 is a configuration block diagram of an automatic equalizer showing an embodiment of the present invention, in which elements common to those in the conventional FIG. 2 are denoted by common reference numerals.

The point that this automatic equalizer is basically different from the conventional one is that a digital processing circuit consisting of a conventional √AGC 5 and a roll-off filter 6 is composed of a variable filter 15 and a fixed filter 16 which are clearly divided into functions. That is. The pre-filter 3 is an LPF of an analog circuit that attenuates frequency components above the clock frequency band of the digital processing circuit.
It is composed of. The digital processing circuit including the variable filter 15 and the fixed filter 16 is configured by using a DSP, a wired logic, or the like.

The variable filter 15 is a filter whose characteristics can be changed according to the line status based on the control signal output from the control circuit 7. For example, when the variable filter 15 is composed of a DSP, the characteristics can be changed by changing the multiplication coefficient thereof, and when the variable filter 15 is composed of a wired logic such as a tapped delay line, The characteristics can be changed by changing the tap weight.

The fixed filter 16 has a constant characteristic regardless of the line condition,
That is, it is composed of a linear phase filter whose phase changes linearly. This linear phase filter is composed of, for example, a one-stage secondary linear phase filter or a plurality of cascaded secondary linear phase filters. The control circuit 7
It has a function of setting the characteristics of the variable filter 15 so that the best equalization can be performed according to the situation of the line. As the adaptive algorithm, for example, a method of previously obtaining the characteristics of the variable filter 15 with respect to some reference lines and selecting the characteristics so that the amplitude value of the output signal So of the equalizer falls within the reference range. Used.

The control circuit 7 is commonly used in this type of automatic equalizer, and a known configuration example thereof is shown in FIG.

The control circuit 7 is a circuit that controls the characteristics of the variable filter 15 based on the output signal So of the fixed filter 16 of FIG. 1, and an amplitude value detection circuit 71 that detects the amplitude value of the output signal So and the amplitude value detection circuit 71. A characteristic setting circuit 72 for setting the characteristic of the variable filter 15 based on the value, and a storage circuit 73 for storing the relationship between the amplitude value and the characteristic of the variable filter 15 capable of the best equalization.

The characteristic of the variable filter 15 should ideally be the inverse characteristic of the line in the signal band. The loss characteristic of the subscriber line depends largely on the line length and not on the line type. Further, if the attenuation amount of the Nyquist frequency is determined, its shape is determined as a so-called √ characteristic. On the other hand, the amount of attenuation at the Nyquist frequency is closely related to the amplitude value of the signal transmitted through the line. Therefore, the optimum characteristic of the variable filter 15 can be determined for the amplitude value of the signal. In the storage circuit 73 in the control circuit 7, for example,
The DSP multiplication coefficient giving the optimum characteristic of the variable filter 15 with respect to the amplitude value of the signal transmitted through the line, the tap weight of the tapped delay line, or the number representing the same is stored. As for the type of characteristic to be stored, the closest characteristic is used when there is no corresponding characteristic, and therefore it is necessary to increase the approximation accuracy to such an extent that it can be practically used.

Next, the operation of the automatic equalizer of FIGS. 1 and 4 configured as above will be described.

In FIG. 1, when the reception signal SI is input to the input terminal 1, the reception signal SI is fed by the pre-filter 3 to, for example, ½ of the clock frequency in the digital processing circuit.
So that the above frequency components are small enough to be ignored,
The signal is band-limited and sent to the A / D converter 4. A /
The D converter 4 samples and quantizes the output signal of the pre-filter 3 and converts it into a digital signal. This digital signal is equalized by the fixed filter 16 and the variable filter 15 set to a desired characteristic by the control signal of the control circuit 7, and the equalized digital output signal So with a small distortion is output terminal 2. And is sent to the control circuit 7.

In the amplitude value detection circuit 71 in the control circuit 7 shown in FIG.
The amplitude value of the output signal So is detected and the amplitude value is sent to the characteristic setting circuit 72. Based on the input amplitude value, the characteristic setting circuit 72 stores, for example, the multiplication coefficient of the DSP giving the optimum characteristic of the variable filter 15, the tap weight of the delay line with taps, or the number representing the same from the storage circuit 73. Read out and set the characteristics of the variable filter 15. This characteristic is determined by the amplitude value detected by the amplitude value detection circuit 71,
This can be realized by converting the difference from the reference amplitude value assumed to be detected in the optimum characteristic into the characteristic difference and changing the characteristic by the difference. That is, for example, in the case where the characteristic is determined for each 0.1 in the amplitude value, if the reference amplitude value is 1 and the detected amplitude value is 0.8, only the two corresponding to the difference 0.2, It suffices to change the characteristics so that the amplitude value increases.

In the above description, the output signal So is assumed to be a digital signal, but if this is D / A converted,
It is also possible to obtain an analog signal.

The feature of this embodiment is that the variable filter 15 and the fixed filter 1 are
The two digital filters of 6 equalize the waveform distortion due to the line loss characteristic in the signal band and are higher than the signal band without adversely affecting the frequency components in the signal band, and the clock frequency band of the digital processing circuit. That is, the two functions of attenuating a frequency component lower than that and removing noise outside the signal band are shared. That is, the function of the variable filter 15 is
Causes each action of to be executed. Then, frequency components above the clock frequency band of the digital processing circuit are attenuated by the pre-filter 3.

In general, the subscriber line has a gentle characteristic that there is no steep phase change, so ignoring the action of
The operation using the relatively simple variable filter 15 can be performed. Especially, for the purpose of canceling the reflected wave of the termination release line called bridged tap, which is usually provided in consideration of the flexibility of the wiring of the subscriber line, a bridged tap etc. is further provided behind this automatic equalizer. A pesticide is used. Therefore, in this automatic equalizer, it suffices that the low-frequency components are mainly equalized, and the operation can be performed by a simple circuit of the second order. Therefore, it can be said that it is advisable to design the fixed filter 16 that performs the function of so as not to interfere with the function of the variable filter 15. At this time, if the fixed filter 16 gives a steep phase change irrespective of the line condition, the design of the variable filter 15 becomes extremely difficult. Therefore, in this embodiment, as the fixed filter 16, a linear phase filter whose phase changes linearly is used.

The transfer function of a simple second-order linear phase filter is H (z) = a + b · z ⁻¹ + a · z ⁻² (1) where a and b are constants. This filter can greatly attenuate the component near the zero frequency determined by the constants a and b, while maintaining the linear phase change. Therefore, by cascading filters of this form with the appropriate zero frequency,
(Clock frequency bandwidth of digital processing circuit) >>
A desired amount of attenuation can be given to a frequency component such as (signal bandwidth). Frequency components above the clock frequency bandwidth of the digital processing circuit are attenuated by the pre-filter 3, so normally this cascade connection is 1
About 2 is sufficient. Therefore, it is possible to operate with a relatively small circuit scale without interfering with the operation of.

3 (a) and 3 (b) are computer simulation results showing the equalization operation of the automatic equalizer in the present embodiment,
FIG. 7A is a solitary wave response diagram after equalization, and FIG. 7B is a frequency characteristic diagram. Incidentally, T in FIG. 3 (a) represents a data identification period.

In Fig. 3, equalization is performed for a line with a wire diameter of 0.5 mm and a length of 4 km, and it is assumed that the transmission speed of the subscriber line is 320 kb / s. The clock frequency of the digital processing circuit is 1.28 MHz, and the pre-filter 3 is a sixth-order non-polar filter. The fixed filter 16 is a second-order linear phase filter represented by the equation (1), and a = 5.728 and b = 9.525. The variable filter 15 has a transfer function of H (z) = N (z) / D (z) ... (2) N (z) = n ₀ + n ₁ z ^-1 + n ₂ z ² (3) ) D (z) = d ₀ + d ₁ · z ⁻¹ + d ₂ · z ⁻² (4) A biquadratic filter is used. Here, the parameters n ₀ , n ₁ , n ₂ , d ₀ , d ₁ and d ₂ are multiplication factors of the DSP or tap weights of the delay line with taps, and are set by the control circuit 7 according to the line conditions. It In the line conditions of this simulation example, n ₀ = −4.645, n ₁ = 14.53, n ₂ = −9.618, d ₀ = 1, d ₁ = −0.15, d ₂ = 0, and Has become. Considering that the bridged type equalizer is connected to the rear side, it is advantageous to slightly over-equalize the output of the line equalizer as described in the above-mentioned reference 2. Also in this embodiment, in consideration of this, the waveform after equalization is slightly over-equalized. Figure 3 (a), (b)
As is clear from the above, even though both the variable filter 15 and the fixed filter 16 are second-order and simple in configuration, equalization of waveform distortion due to line loss characteristics within the signal band and noise removal outside the signal band are performed. It can be seen that the above two effects are satisfactorily realized.

The present invention is not limited to the illustrated embodiment, and various modifications can be made. For example, in the above-described embodiment, when the fixed filter 16 is configured, a high-order linear phase filter is configured by cascading the second-order linear phase filters as a unit, but the high-order linear phase filter is directly connected. Even if the linear phase filter is configured, the same operation as the above embodiment,
The effect is obtained.

[Advantage of Device] As described in detail above, according to the present invention, the suppression of the frequency component above the clock frequency band of the digital processing circuit is performed by the pre-filter, and the compensation of the line characteristic in the signal band is performed by the variable filter. As the fixed filter suppresses the frequency components above the signal band and below the clock frequency band of the digital processing circuit, the three filters clearly share the functions. Further, the fixed filter does not adversely affect the signal band. Since it is composed of a linear phase filter which is difficult to give, this kind of automatic equalizer can be appropriately realized with a relatively small circuit scale. Moreover, since the functions are clearly assigned to each circuit, the effect of facilitating the design can be expected.

[Brief description of drawings]

FIG. 1 is a block diagram of an automatic equalizer showing an embodiment of the present invention, FIG. 2 is a block diagram of a conventional automatic equalizer, and FIGS. 3 (a) and 3 (b) are in FIG. FIG. 4A is a solitary wave response diagram, FIG. 4B is a frequency characteristic diagram, and FIG. 4 is a configuration block diagram of a control circuit. 3 ... pre-filter, 4 ... A / D converter, 7 ... control circuit, 15 ... variable filter, 16 ... fixed filter,
71 ... Amplitude value detection circuit, 72 ... Characteristic setting circuit, 73
...... Memory circuit.

Claims (2)

[Scope of utility model registration request] 1. A prefilter configured by a low-pass filter of an analog circuit for band limiting an input signal, an analog / digital converter for converting an analog signal output from the prefilter into a digital signal, and the analog / digital converter. An automatic digital subscriber line etc. comprising a digital processing circuit for equalizing the output signal of the converter and a control circuit for controlling the characteristics of the digital processing circuit based on the amplitude value of the output signal of the digital processing circuit. In the digitalizer, the pre-filter is configured to attenuate frequency components above a clock frequency band of the digital processing circuit, and the digital processing circuit is variably controlled in characteristics by the control circuit according to a line state,
A variable filter for equalizing the line loss characteristic in the signal band with respect to the output signal of the analog / digital converter, and a linear phase filter having a constant characteristic irrespective of the line condition, which is higher than the signal band and An automatic equalizer characterized by comprising a fixed filter that attenuates frequency components lower than the clock frequency band. 2. As the fixed filter, a second-order linear phase filter whose transfer function is represented by H (Z) = a + bZ ^-1 + aZ ^-2 , a, b: constant is used, or the second-order linear phase filter is used. The automatic equalizer according to claim 1, wherein a plurality of linear phase filters are connected in cascade and used.