JP2001189052A - Waveform equalizer - Google Patents

Abstract

PROBLEM TO BE SOLVED: To realize a PR4 (1, 0, -1) equalizer providing a precise tap coefficient even where an S/N ratio is low. SOLUTION: A tap coefficient decision part 622 defines a difference between a PR4 teacher signal series 661 performing 1+D operation 660 and the output 681 of the PR4 (1, 0, -1) equalizer as an equalization error, and further, updates an amount decided considering a size of an equalizer input 671 performing 1+D operation 670, and the tap coefficient of a convolution circuit 521 of PR4 equalizer 405 so that the equalization error becomes smaller.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a waveform equalizer having a PR4 (1, 0, -1) response characteristic.

[0002]

2. Description of the Related Art In recording devices such as a magnetic disk device (for example, a magnetic disk device, a magnetic tape device) and an optical or magneto-optical disk device (for example, an optical disk device, a magneto-optical disk device), a partial response (PR) is used.
onse) method to perform waveform equalization and detect data with maximum likelihood.
A PRML method performed by a (ML: Maximum Likelihood) decoding method has been put to practical use. In this PRML method, data is reproduced by actively utilizing intersymbol interference (interference of reproduced signals corresponding to adjacent recording bits), and the most reliable data is obtained by using data of a sample point and sample points before and after it. It is assumed that a likely data series is decoded data. By using the PRML system, the band of the signal channel is limited, and as a result, the detection accuracy can be improved even for a reproduced waveform having a low S / N ratio, so that the recording density in the recording apparatus can be increased. Here, for maximum likelihood decoding, a Viterbi detector using a Viterbi algorithm is generally used. On the other hand, there are several PR systems depending on what kind of intersymbol interference is applied. For example, in a magnetic disk drive, a PR4 (1, 0, -1) system having band-pass characteristics is often used.
Also, recently, EPR (Extended
PR: 1, 1, -1, -1) or EEPR (Extended EPR: 1, 2,
The 0, -2, -1) method is also attracting attention. PR, EPR, the transfer function of each type of EEPR, respectively G (D) = 1-D 2, G (D) = (1-
D ² ) (1 + D), G (D) = (1-D ² ) (1 + D) ² and the length (constraint length) L of the code interference is 2 for the PR method and 2 for the EPR method. 3 、 EEPR method is 4
Becomes Now, the signal whose waveform has been equalized in each of the PR systems has a correlation with the preceding and following signals. Therefore, the maximum likelihood decoding is performed by utilizing the characteristics of the signal based on the amplitude correlation by the PR method.
The PRML method estimates the most likely value and performs decoding. FIG. 10 shows a typical configuration of a signal processing system of a conventional magnetic disk device employing such a PRML system.
In the figure, the data input to the signal processing circuit unit 420 is
The data is converted into a specific code in the encoder 401 and input to the precoder 402, where the precoder performs data modulation processing. The data to be recorded is binary data and takes a value of "0" or "1". The R / W amp 410 generates a signal that is high when the output data of the precoder 402 is "1" and low when the output data is "0".
Data is recorded on the head via the head 412. The signal read from the magnetic disk medium 411 via the head 412 is R / W
An analog filter 403 that removes high-frequency noise superimposed on the signal amplified by the amp 410, passes through an A / D converter 404 that performs analog / digital conversion of the signal at predetermined time intervals, and is output by the waveform equalizer 405. Are equalized so as to have a PR4 (1, 0, -1) response waveform. A reproduced signal in the case of magnetic recording is obtained by performing a 1-D operation on data on the medium from the characteristics of the head. Here, D is a delay operator. The signal equalized by the waveform equalizer 405 passes through a channel 406 having a specific response characteristic, and then to a channel in which all the PR4 (1, 0, -1) characteristics and the specific response characteristic are considered. It is input to a maximum likelihood decoder (eg, Viterbi detector) 407 that is configured to fit. The maximum likelihood decoder 407 performs maximum likelihood estimation, and sends the estimation result to the decoder 408. The decoder 408 restores and outputs an input data sequence to the signal processing circuit 420. As a recording apparatus having such a configuration, for example, Japanese Patent Application Laid-Open
The device described in 58514 is known. Next, a conventional waveform equalizer 405 used in such a conventional magnetic disk device will be described. As shown in FIG. 11, a conventional waveform equalizer 405 includes a convolution circuit 521 and a tap coefficient determination unit 522.
Consists of The convolution circuit 521 has N taps numbered from 0 to N-1,

[0003]

(Equation 1)

[0004] This is an FIR filter that performs an output _Pn according to the above.
Here, h ⁱ _n the tap coefficients of the equalizer, x _n respectively represent the input to the equalizer. In P, h, x, a subscript indicates a time, and a superscript indicates a tap number.

[0005] Further, a tap coefficient determination unit 522 calculates a tap coefficient h of each tap from the teacher signal 530, the A / D converter output 511, and the equalizer output 532.

[0006]

(Equation 2)

This is a circuit which is determined according to the following. Here, the T _n teacher signal, e _n is the equalization error, alpha represents respectively the acceleration factor. The subscripts in T and e indicate time.

Here, the teacher signal 530 is a target value of the equalizer output 523 to be obtained for the A / D converter output 511, which is prepared in advance for the known A / D converter output 511. A PR4 (1,
0, -1) is a signal sequence. The acceleration coefficient α is a known LMS
(Least Mean Square) is a coefficient that is calculated statistically by an algorithm or the like. Note that in practice,
The values of e _n and x _n are usually averaged and returned at appropriate times for stabilization. Therefore, the update of the tap coefficient is performed at intervals of this time. Now, the waveform equalizer 405 configured as above converges to output an input equalized to PR4 (1, 0, -1). In the following, such PR4 (1, 0,
The waveform equalizer that performs the output equalized to -1) is simply called a PR4 equalizer.

[0009]

The PR4 shown in FIG.
Due to the characteristics of the equalizer, there is a problem that the tap coefficient value for performing accurate output due to the influence of noise, particularly the output is difficult to find, especially at a high S / N ratio (signal-to-noise ratio). Therefore, the present invention provides a more accurate tap coefficient value,
Therefore, it is an object of the present invention to provide a PR4 (1, 0, -1) equalizer that can obtain an output.

[0010]

In order to achieve the above object,
The present invention provides, for example, a method of updating a value of the finite number of coefficient sequences in an equalizer that outputs a predetermined partial response characteristic as a target response characteristic by convolving an input data sequence with a finite number of coefficient sequences. The first sequence subjected to an operation of giving a predetermined inter-symbol interference to a target output data sequence for the input data sequence of the equalizer and the predetermined inter-symbol interference to the output sequence of the equalizer Obtaining an equalization error that is a difference from the second sequence that has been subjected to the operation to be given, and a third sequence that has been subjected to the operation that gives the predetermined inter-symbol interference to the input data sequence; and So as to reduce the amount according to the equalization error and the third sequence, and a step of updating the value of a finite number of coefficient sequences of the equalization means. A method for updating coefficients of an equalizer is provided. According to such a method, by using an appropriate operation as an operation for giving a predetermined intersymbol interference, a more appropriate coefficient value, and hence an equalizer You will be able to get output.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described by taking an application to a magnetic disk drive as an example. The configuration of the signal processing system of the magnetic disk device of the magnetic disk device according to the present embodiment is the same as the configuration shown in FIG. However, the internal configuration of the waveform equalizer 405 is as shown in FIG.
Are different from those shown in the above. FIG. 1 shows an internal configuration of a waveform equalizer (PR4 equalizer) 405 according to the present embodiment. As shown in the figure, in the waveform equalizer 405 according to the present embodiment, the tap coefficient determination unit 622 outputs 1 + D to the A / D converter output 511 by the 1 + D calculator 660.
A teacher signal (hereinafter, referred to as an “EPR teacher signal”) obtained by performing 1 + D operation on the ADC output 671 subjected to the operation and the same teacher signal 550 as the conventional one by the 1 + D calculator 670 to obtain an EPR signal sequence. 661) and the signal sequence 681 obtained by performing 1 + D operation on the equalizer output 623 by the 1 + D operation unit 680, from 0 to N-1 of the convolution circuit 521 same as the conventional one.
The tap coefficient values of N taps given numbers up to are determined. The tap coefficient value is determined, for example, according to the following equation.

[0012]

(Equation 3)

[0013] In Equation 3, the P _n is the equalizer output, the tap coefficients of h ⁱ _n is the equalizer, the input to x _n is the equalizer, the T _n teacher signal, e _n is equal And α is an acceleration coefficient. In P, h, x, T and e, the subscripts indicate the time and h
The superscript suffix indicates the tap number.

Here, the teacher signal 530 is a target value of the equalizer output 623 to be obtained for the A / D converter output 511, which is prepared in advance for the known A / D converter output 511. A PR4 (1,
0, -1) is a signal sequence. The acceleration coefficient α is determined in advance by LMS (Lea
This is a coefficient that is obtained statistically by the st mean square (algorithm) algorithm or the like. Note that, in practice, the quantities e _n , x
The value of _n is averaged and fed back for an appropriate time, usually for stabilization. Therefore, the update of the tap coefficient is performed at intervals of this time. Now, when updating the tap coefficients in the direction of reducing the equalization error, the more noise superimposed on the AD converter output 511, the smaller the absolute value of the amplitude of the tap coefficients generally obtained. As a result, the equalization accuracy is deteriorated. This is because reducing the absolute value of the tap coefficient corresponds to reducing the apparent equalization error. However, as in the present embodiment, the equalizer output 623, the PR4 teacher signal 550, and the AD converter output 511 have (1
+ D) When the tap coefficient value of the PR4 equalizer is determined from the sequence subjected to the operation, the tap coefficient value can be obtained with higher accuracy than when the tap coefficient value is determined by the conventional method as shown in FIG. . The reason is described below. First, Fig. 2 shows PR4 (800), EPR
(801) and frequency characteristics of each channel of EEPR (802) are shown. As can be seen from this figure, the frequency characteristics of the EPR channel and the EEPR channel have a higher attenuation rate in the high frequency band than the frequency characteristics of the PR4 channel. When used for an EPR teacher signal, which is an EPR signal train, higher frequencies can be attenuated due to channel characteristics as compared with the case where a conventional teacher signal of a PR4 signal train (hereinafter, referred to as “PR4 teacher signal”) is used. Since the tap coefficients are updated under such a premise, the tap coefficient value obtained by using the EPR teacher signal is smaller in the attenuation rate of the high-frequency signal (larger amplitude). Can be obtained. FIGS. 3 and 4 show examples of tap coefficient values of the PR4 equalizer when the conventional PR4 teacher signal is used and when the EPR teacher signal is used. FIG. 4 shows an example of the obtained tap coefficient values, and FIG. 9 shows the frequency characteristics thereof. In both figures, the line of the PR teacher signal (900, 1000) represents the characteristic of the tap coefficient of the conventional PR4 equalizer, whereas the line of the EPR teacher signal (901, 1001) is according to the present embodiment. 9 shows characteristics of a tap coefficient of a PR4 equalizer. From this figure, it can be seen that a tap coefficient with a lower attenuation rate in the high frequency range can be obtained when the EPR teacher signal is used than when the PR4 teacher signal is used. Then, as shown in a graph in which the S / N ratio is plotted on the horizontal axis and the bit error rate is plotted on the vertical axis in FIG. 5, such a PR4 equalizer according to the present embodiment is
The bit error rate (1101) at the maximum likelihood decoder output when the signal sequence PR4 equalized by 405 is decoded by the maximum likelihood decoding decoder 407 through an appropriate partial response channel 406 is the conventional PR4
The signal sequence PR4 equalized by the equalizer, the bit error rate (1100) at the output of the maximum likelihood decoder when decoded by the maximum likelihood decoding decoder 407 through an appropriate partial response channel 406, can be kept low. become. As above, the first embodiment of the present invention has been described. Hereinafter, a second embodiment of the present invention will be described. The configuration of the signal processing system of the magnetic disk device of the magnetic disk device according to the present embodiment is the same as the configuration shown in FIG. However, the internal configuration of the waveform equalizer 405 is as shown in FIG.
Are different from those shown in the above. FIG. 6 shows an internal configuration of the waveform equalizer (PR4 equalizer) 405 according to the present embodiment. As shown in the figure, in the waveform equalizer 405 according to the present embodiment, the tap coefficient determination unit 722 outputs the A / D converter output 511 to the (1 + D) ² arithmetic unit 760.
(1 + D) ² ADC output 771 with operation and 5 teacher signals as before
A teacher signal (hereinafter, referred to as an “EEPR teacher signal”) 761 obtained by performing (1 + D) ² operations on 50 with (1 + D) ² arithmetic unit 770, and an equalizer (1 + D) ² for output 623
A tap coefficient value of N taps given numbers from 0 to N−1 of the convolution circuit 521 is determined from the signal sequence 781 subjected to (1 + D) ² operation in 80. The tap coefficient value is determined, for example, according to the following equation.

[0015]

(Equation 4)

In equation (4), P _n denotes the output of the equalizer, h ⁱ _n
Denotes a tap coefficient of the equalizer, x _n denotes an input to the equalizer, T _n denotes a teacher signal, en _denotes an equalization error, and α denotes an acceleration coefficient. In P, h, x, T and e, the subscripts indicate the time, and the superscripts indicate the tap numbers.

Here, the teacher signal 530 is a target value of the equalizer output 623 to be obtained for the A / D converter output 511, which is prepared in advance for the known A / D converter output 511. A PR4 (1,
0, -1) is a signal sequence. The acceleration coefficient α is determined in advance by LMS (Lea
This is a coefficient that is obtained statistically by the st mean square (algorithm) algorithm or the like. Note that, in practice, the quantity e _n ,
The value of x _n is averaged and fed back for an appropriate time, usually for stabilization. Therefore, the update of the tap coefficient is performed at intervals of this time. Even when the EEPR teacher signal is used in this manner, as shown in FIG. 2, the frequency characteristic of the EEPR channel has a higher attenuation rate in the high frequency band as compared with the frequency characteristic of the PR4 channel. In addition, the tap coefficient value can be obtained with higher accuracy than when the tap coefficient value is determined by the conventional method shown in FIG. Note that, in FIGS. 3 and 4, the EEPR teacher signal (902, 1002) shows the characteristic of the tap coefficient obtained by the second embodiment, and 1102 in FIG. 5 shows the bit error rate according to the second embodiment. It is. The first and second embodiments have been described above. Note that the maximum likelihood decoder 407 in the first and second embodiments compares an appropriate threshold with an input value,
The comparison result may be used as the decoding result. Also,
The signal processing system of the magnetic disk device according to the first and second embodiments can be used, for example, in the magnetic disk device shown in FIG. 7 includes a magnetic disk 100 on which data is written, a spindle motor 101 for rotating the disk 100, a head 103 for reading data from the disk 201, and an arm 10 supporting the head 103.
2, voice coil motor 10 for moving head 103
4. The R / W amp 105 amplifies the signal from the head 103. In addition, this magnetic disk device, as an electronic circuit portion,
As shown in FIG. 8, an interface 220 for connecting to an information processing device such as a host, an interface control circuit 210 for controlling input / output of the interface 220, a magnetic disk device controller 211 for controlling data transfer and format, etc. , A microprocessor 212 and a spindle control circuit 2 for controlling the spindle motor 101
14, voice coil motor control circuit 213, signal processing circuit 2
With 15. R / W amp 410 of the signal processing system shown in Fig. 11
However, the R / W amp 105 of FIG. 7, the signal processing circuit unit 420 of the signal processing system shown in FIG. 11 is used for the signal processing circuit 215 of FIG. 8, and the magnetic disk medium 411 shown in FIG. In the disk 100, the head 412 shown in FIG. 11 corresponds to the head 103 in FIG. Also figure
The magnetic disk devices shown in FIGS. 7 and 8 can be used, for example, to configure an information processing system as shown in FIG. As shown in FIG. 9, the information processing device 300 includes a CPU 301, a memory 302, a bridge 303, and a peripheral interface 304. The CPU 301 and the memory 302 are connected by an internal bus 305. The peripheral interface 304 is connected to an external bus 306, and the internal bus 305 and the external bus 306 are connected by a bridge 303. The information processing device 300 connects to an external peripheral device through the peripheral interface 304. The magnetic disk device shown in FIGS. 7 and 8 is connected as this external peripheral device. The information processing device 300 reads and writes data on the magnetic disk 100 via the peripheral interface 304 and the interface 310 in the magnetic disk. The embodiment of the invention has been described. Note that the waveform equalizer (PR4 equalizer) according to the present embodiment is not only a magnetic disk device,
The present invention can be similarly applied as a PR4 equalizer for various storage devices and other devices.

[0018]

As described above, according to the present invention, it is possible to provide a PR4 (1, 0, -1) equalizer capable of obtaining a more accurate tap coefficient value and, therefore, an output. .

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a waveform equalizer (PR4 equalizer) according to a first embodiment of the present invention.

FIG. 2 is a diagram showing frequency characteristics of each channel of PR4 (1, 0, -1), EPR, and EEPR.

FIG. 3 is a diagram showing amplitude values of tap coefficients according to the related art and the first and second embodiments.

FIG. 4 is a diagram illustrating frequency characteristics of tap coefficients according to the related art and the first and second embodiments.

FIG. 5 is a diagram showing a bit error rate according to the conventional and first and second embodiments.

FIG. 6 is a block diagram illustrating a configuration of a waveform equalizer (PR4 equalizer) according to the first embodiment of the present invention.

FIG. 7 is a block diagram showing a configuration of a magnetic disk drive according to an embodiment of the present invention.

FIG. 8 is a block diagram showing a configuration of a magnetic disk drive according to an embodiment of the present invention.

FIG. 9 is a block diagram showing a configuration of an information processing system according to an embodiment of the present invention.

FIG. 10 is a block diagram illustrating a configuration of a signal processing system according to the related art and each embodiment of the present invention.

FIG. 11 is a block diagram showing a configuration of a conventional waveform equalizer (PR4 equalizer).

[Explanation of symbols]

 401: encoder 402: precoder 403: analog filter 404: A / D converter 405: PR4 waveform equalizer 406: partial response channel 407: maximum likelihood detector 408: decoder 410: R / W amp 411: magnetic disk 420: Signal processing circuit section 500: Analog filter 510: Analog / digital converter 520: PR4 (1, 0, -1) equalizer 521: Convolution circuit 522: Tap coefficient determination section 530: Partial response channel 540: Maximum likelihood decoder 622: Tap coefficient determination unit 660, 670 680: 1 + D calculator 722: Tap coefficient determination unit 760, 770, 780: (1 + D) 2 calculator

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Takushi Nishitani 1099 Ozenji Temple, Aso-ku, Kawasaki City, Kanagawa Prefecture Inside of Hitachi, Ltd.System Development Laboratory (72) Inventor Nobuaki Nakai 5--20, Josuihoncho, Kodaira-shi, Tokyo No. 1 In the Hitachi, Ltd. Semiconductor Group (72) Inventor Hiroshi Ide 5-2-1, Kamimihoncho, Kodaira-shi, Tokyo F-term in the Hitachi, Ltd. Semiconductor Group 5D044 BC01 CC04 FG02 FG05 FG30

Claims (6)

[Claims] 1. A waveform equalizer having a partial response class 4 (1, 0, -1) response characteristic as a target response characteristic, wherein an equalization is performed by convolving an input data sequence with a finite number of coefficient sequences and outputting the convolution. Means,
Coefficient equalizing means for updating the value of a finite number of coefficient sequences of the equalizing means, wherein the coefficient determining means sets a target output for the input data sequence of the equalizing means so as to reduce the equalization error. A first sequence obtained by performing (1 + D) operation on the data sequence and a second sequence obtained by performing (1 + D) operation on the output sequence of the equalization means
An equalization error that is a difference from the sequence of the amount, an amount corresponding to a third sequence that has been subjected to the (1 + D) operation on the input data sequence, and a value of a finite number of coefficient sequences of the equalization means. A waveform equalizer characterized by updating. 2. A waveform equalizer having a partial response class 4 (1, 0, -1) response characteristic as a target response characteristic, wherein the equalization is performed by convolving an input data sequence with a finite number of coefficient sequences and outputting the convolution. Means,
Coefficient equalizing means for updating the value of a finite number of coefficient sequences of the equalizing means, wherein the coefficient determining means sets a target output for the input data sequence of the equalizing means so as to reduce the equalization error. The first of (1 + D) ² operations on the data series
Facilities and equalization error, the input data series (1 + D) ² operation sequence and is the difference between the second series subjected to (1 + D) ² operation on the output line of said equalizing means A waveform equalizer characterized by updating an amount corresponding to the third sequence and a value of a finite number of coefficient sequences of the equalizing means. 3. A waveform equalizer according to claim 1 or 2, an analog / digital converter for converting an input signal from analog to digital to obtain an input data sequence of said waveform equalizer, and said waveform equalizer. And a maximum likelihood decoder for maximum likelihood decoding of the output sequence. 4. A recording apparatus comprising: the signal processing circuit according to claim 3; a storage medium; and a reading unit that reads a signal stored in the storage medium and uses the read signal as an input of the signal processing circuit. . 5. The recording apparatus according to claim 3, wherein said recording medium is a magnetic disk. 6. A predetermined partial response characteristic is set as a target response characteristic,
An equalizer that convolves and outputs an input data sequence with a finite number of coefficient sequences, and a method for updating the value of the finite number of coefficient sequences, wherein a target output data sequence for the input data sequence of the equalizer is An equalization error, which is a difference between a first sequence subjected to an operation of giving predetermined intersymbol interference and a second sequence subjected to an operation of giving the predetermined intersymbol interference to an output sequence of the equalizer. And a step of obtaining a third sequence that has been subjected to the operation of giving the predetermined inter-symbol interference to the input data sequence, and so that the equalization error is reduced, the equalization error and the third sequence Updating a value of a finite number of coefficient sequences of the equalizer according to a corresponding amount, the coefficient updating method of an equalizer having partial response characteristics.