JP3087314B2 - Adaptive filter - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION The present invention relates to adaptive filters.
In particular, the present invention relates to an adaptive filter for removing an interference component from a signal in which intersymbol interference has occurred.

[0002]

2. Description of the Related Art In a recording / reproducing system such as an optical disk, a run length limited code in which a minimum inversion interval and a maximum inversion interval of a code are limited is often used. For example, (2,7) having the (2,7) RLLC encoding table shown in Table 1
RLLC is a modulation code in which the run length of “1” is limited to 1 and the run lens of “0” is limited to 2 to 7. Here, furthermore, by performing NRZI recording with “1” corresponding to the inversion of the code, the same code becomes a run-length limited code such that the same code continues from 3 bits to 8 bats. Also,
The compact disk employs NRZI recording of the EFM modulation, and the same code appears continuously from 3 bits to 11 bits in the reproduced signal.

[0003]

In general, it is difficult to form a recording bit having a size smaller than a certain limit on an optical disk. Therefore, the use of a code having a large minimum code inversion interval can limit the lower limit of the recording bit size. There is. Further, since the frequency band of the recording code can be limited, the recording code may be used for magnetic recording such as a magnetic tape or a magnetic disk.

[0005] Even when a run-length limited code is used,
When the recording density is high to some extent, it becomes difficult to restore correct data due to interference according to the pattern of the reproduction code. In such a case, in order to remove the waveform distortion generated between the codes, the equalizer is provided in the reproduction circuit to compensate for the intersymbol interference in the signal sequence. When the transfer characteristics in the recording / reproducing system change or when the transfer characteristics cannot be specified, a method of adaptive equalization is adopted in which waveform distortion is estimated from the reproduced signal to determine the characteristics of the equalizer.

FIG. 5 is a block diagram showing a configuration example of a conventional adaptive filter. The signal before equalization input from the input terminal 1 is sampled by the sampling circuit 2 at a sample rate equal to the channel bit rate. The sampled signal has its waveform distortion removed by a transversal filter 3 including a delay element 10, a multiplier 11, and an adder 12, and is output from an output terminal. At the same time, this output signal is input to an error signal extraction circuit 6 to extract an error signal used for updating the tap gain.

If the tap gain is not set to an optimum value, a signal with waveform distortion is output from the output terminal. The decision unit 15 performs a binary decision on the signal with the remaining waveform distortion, and outputs a binary signal having a reference amplitude. The tap gain coefficient changes in a direction to minimize the power of the error signal obtained by the difference between the output signal of the transversal filter 3 and the output signal of the decision unit 15. For this reason, after the tap coefficient converges to an appropriate value, the filter output signal takes the two values given as the output of the determiner.

The error signal is supplied to a correlator 20 together with the input signal of a suitable time-delayed filter provided as a reference signal.
And the correlator 20 outputs the correlation strength of both. The output signal of the correlator 20 is integrated by the integrator 21 and supplied to the multiplier 11 of each tap as a tap gain coefficient.

[0009]

In a recording / reproducing system requiring a high data transfer rate, high-speed processing is required in a reproducing signal processing circuit. However, the above-mentioned conventional adaptive filter requires an operation speed equal to the channel bit rate, and it has been difficult to realize a circuit capable of supporting a high transmission speed.

An object of the present invention is to solve the above problems,
An object of the present invention is to provide an adaptive filter that operates well at an operating frequency of 1/2 or less of a channel bit rate.

[0011]

SUMMARY OF THE INVENTION An adaptive filter according to the present invention is an adaptive filter for equalizing a signal having a waveform distortion modulated by a run-length limited code. A sampling circuit for sampling at a frequency equal to or lower than / 2, a transversal filter having a variable tap gain for inputting an output signal of the sampling circuit and changing the frequency characteristic and outputting the same, and an output signal of the transversal filter for inputting. A first interpolation circuit that outputs a signal at a data rate equal to a channel bit rate, and an error extraction circuit that receives an output signal of the first interpolation circuit and outputs a difference between the amplitude level as a reference and a determination error signal A sign inversion detection circuit for receiving an output signal of the first interpolation circuit and detecting a sign inversion position; A second interpolation circuit for outputting a signal with data rate equal to the channel bit rate by inputting the output signal of the sampling circuit, the second interpolation circuit,
Mutual signals corresponding to channel bits immediately before and immediately after the sign inversion position in the output signal of the second interpolation circuit and the output signal of the error extraction circuit by inputting the output signals of the error extraction circuit and the sign inversion detection circuit. And a coefficient control circuit for controlling the tap gain of the transversal filter by the correlation of

[0012]

When equalizing a signal sampled at a sampling rate equal to or less than 1/2 of the channel bit rate by a transversal filter, equalizing so that all codes of the channel bits take a specific binary level is not possible. Can not. However, the run-length limited code can be decoded if the inversion position of the code can be specified.

In the present invention, a method is used in which a signal obtained by interpolating an output signal of a transversal filter is partially equalized so that only bits before and after sign inversion take a specific binary level. As a result, an adaptive filter that can accurately determine the reproduction data even at a low operating frequency can be realized.

[0014]

Next, an embodiment of the present invention will be described with reference to FIGS.

FIG. 1 is a block diagram showing a first embodiment of the adaptive filter according to the present invention. The signal input from the input terminal 1 is sampled by the sampling circuit 2 at a frequency of a channel bit rate １ ／. The sampled signal is input to a variable tap gain transversal filter 3 composed of a delay element 10 having a delay amount of 2T, a multiplier 11 and an adder 12, and is output after changing its frequency characteristic. Here, T is the time occupied by one channel bit. The signals output from the torsional filter 3 at 2T intervals are interpolated by the first interpolation circuit 4 into signals at time T intervals.

The output signal of the first interpolation circuit 4 is subjected to binary decision by a decision unit 15 provided in the error extraction circuit 6, and a difference from a reference level is output as an error signal. At the same time, the output signal of the first interpolation circuit 4 is
Control signal is activated when bits immediately before and after sign inversion appear. Further, the second interpolation circuit 8 interpolates the input signal sampled at a time interval of 2T into a signal at a time interval of T and outputs it as a reference signal for coefficient control.

The tap coefficient control circuit 9 includes a delay element 10, a switch 19, a correlator 20, and an integrator 21. The switch 19 is closed when the control signal input from the sign inversion detection circuit 7 is active, and among the error signals output from the error signal extraction circuit 6, only the error signal corresponding to the bits immediately before and immediately after the sign inversion. Is input to the correlator 20.

Each correlator 20 outputs the product of the delayed reference signal and the error signal. Further, the output of the correlator 20 is integrated by the integrator 21 and sent to the transversal filter 3 as a tap gain coefficient of the transversal filter 3. When switch 19 is open,
Since the error signal input to the correlator 20 is 0, the tap gain coefficient does not change.

FIG. 2 is a diagram showing waveforms at various parts in FIG. 1. The operation of this embodiment will be described with reference to FIGS. FIG. 2A shows an example of recording data encoded by a run-length control code. The same code appears at least three consecutive times. The time occupied by one channel bit is T. The reproduced signal is obtained with waveform distortion as shown in FIG. In the case of a magnetic tape or a magnetic disk, a reproduced signal is obtained as a differential waveform of the waveform b, but a waveform similar to the waveform b can be easily obtained by giving the reproducing amplifier an integral characteristic.

FIG. 2C shows an input signal of the transversal filter 3 sampled by the sampling circuit 2, and FIG. 2D shows an output signal of the transversal filter 3. Since both the input and output signals of the transversal filter 3 are handled as signals with a sampling interval of 2T, the processing speed of the filter is 1 / 1.5 as compared with the conventional example.
It becomes 2. The output signal d of the transversal filter 3 is interpolated at intervals of T by the interpolation circuit 4 as shown in FIG. The amplitude of the signal obtained by interpolation takes the middle point of the output signal d sequence. The error extraction circuit 6 performs a binary data decision from the amplitude of the output signal e column and outputs an error signal f.

FIG. 2g shows a control signal output from the sign inversion detection circuit 7. The sign inversion detection circuit 7 detects a sign change of the output signal e of the first interpolation circuit 4, and activates the control signal g for a time 2T immediately after the sign change. The switch 19 provided in the coefficient control circuit 9 is closed when the control signal g is active, and sends the error signal f to the correlator 20. The error signal sent to the switch 19 is the error extraction circuit 6
Since the output signal f is delayed by the time T, it corresponds to the channel bit signal immediately before and after the sign inversion. Correlator 2
A signal obtained by interpolating the sampled input signal c is input to 0 along with the error signal. The tap gain coefficient is controlled by the correlation between them as in the conventional example.

FIG. 3 is a block diagram showing a second embodiment of the adaptive filter according to the present invention. The error extraction circuit 6, the sign inversion detection circuit 7A and the tap coefficient control circuit 9 are each configured as a parallel circuit. The first and second interpolation circuits 4A,
8A is obtained by removing the parallel / serial converter 14 from the circuits 4 and 8 having the same name in FIG. In this configuration, a signal interpolated at time T intervals is divided into two parallel signals and processed. Therefore, not only the operating frequency of the transversal filter 3 but also the operating frequencies of other circuits can be suppressed to 1 / (2T) or less.

The structure of the transversal filter 3 is shown in FIG.
In the same manner as described above, the input signal sampled at 2T intervals is filtered and output. Of the two outputs of the first interpolation circuit 4A, the signal output from the output terminal 5-1 corresponds to the signal preceding the output signal from the output terminal 5-2 by the time T. The error signal extraction circuit 6A is constituted by the subtracters 16 of the two determiners 15, and each outputs an error signal of each system independently.

The sign inversion detecting circuit 7A detects the sign inversion position from the interpolated signal input as a parallel signal, and operates so as to activate the control signal with bits immediately before and after the sign inversion. However, of the two signals after the interpolation, one of the signals corresponding to the output signal from the output terminal 5-2 has a time delay of 2T in detecting the sign inversion.

The tap coefficient control circuit 9A is provided with two systems of correlators 20, and the two parallel reference signals and error signals are processed in parallel. However, one of the error signal and the reference signal is provided with a delay element 10 for compensating for a delay in detecting the sign inversion. Correlator 2
The switches 19-1 and 19-2 provided at the input of 0 are closed only when the control signal input from the sign inversion detection circuit 7A is active, similarly to the switch circuit 19 of FIG. The outputs of the correlators 20 also provided in parallel are added and integrated for each corresponding tap, and are provided to the transversal filter 3 as tap gain coefficients.

FIG. 4 is a block diagram showing another example of the interpolation circuit in FIG. In addition to the configuration of the interpolating circuits 4A and 8A shown in FIG. 3 for interpolating using the signals of two sample points immediately before and after, a configuration for interpolating from four sample points as shown in FIG. Can also. At this time, the coefficients of the constant coefficient multiplier 13 are set to 2 / π and 2 /
When set to (3π), 1 / contained in the interpolated signal
Since the frequency component of (4T) or more can be suppressed small, high equalization accuracy can be obtained. However, the delay time of the output signal increases by 2T.

[0027]

As described above, according to the present invention, it is possible to obtain a higher data transfer rate than before without increasing the operating frequency of the adaptive filter.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a first embodiment of an adaptive filter according to the present invention.

FIG. 2 is a diagram showing waveforms at various parts in FIG.

FIG. 3 is a block diagram showing a second embodiment of the adaptive filter according to the present invention.

FIG. 4 is a block diagram showing another example of the interpolation circuit in FIG. 3;

FIG. 5 is a block diagram illustrating a configuration example of a conventional adaptive filter.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Input terminal 2 Sampling circuit 3 Transversal filter 4, 4A First interpolation circuit 5, 5-1, 5-2 Output terminal 6, 6A Error extraction circuit 7, 7A Sign inversion detection circuit 8, 8A Second interpolation circuit 9, 9A Tap coefficient control circuit 11 Multiplier 13 Constant coefficient multiplier 14 Parallel / serial converter 15 Judge 19, 19-1, 19-2 Switch 20 Correlator 21 Integrator

──────────────────────────────────────────────────の Continued on the front page (51) Int.Cl. ⁷ identification code FI H03H 21/00 H03H 21/00 H04B 3/06 H04B 3/06 D (58) Investigated field (Int.Cl. ⁷ , DB name) ) H03H 17/00 601 G11B 7/00 G11B 20/10 321 H03H 15/00 H03H 17/06 635 H03H 21/00 H04B 3/06

Claims (1)

(57) [Claims] 1. An adaptive filter for equalizing a signal having waveform distortion modulated by a run-length limited code, wherein an input signal is １ ／ of a channel bit rate.
A sampling circuit that samples at the following frequencies,
A transversal filter having a variable tap gain for inputting an output signal of the sampling circuit and changing and outputting a frequency characteristic, and inputting an output signal of the transversal filter and outputting a signal at a data rate equal to a channel bit rate. A first interpolation circuit, an error extraction circuit that receives an output signal of the first interpolation circuit and outputs a difference between the amplitude level as a reference as a determination error signal, and an output signal of the first interpolation circuit. A sign inversion detection circuit for inputting and detecting a sign inversion position, a second interpolation circuit for receiving an output signal of the sampling circuit and outputting a signal at a data rate equal to a channel bit rate, and the second interpolation circuit , An output signal of the error extraction circuit and the sign inversion detection circuit, and an output signal of the second interpolation circuit and the error extraction circuit. Adaptive filter, characterized in that it comprises a coefficient control circuit for controlling the tap gain of the transversal filter by the correlation of the immediately preceding and mutual signal corresponding to the channel bit immediately following the sign inversion position of the force signal.