JPH05290525A - Magnetic disk device - Google Patents

Abstract

PURPOSE:To reduce an erroneous decision in Viterbi equalization even if S/N is low by coding recording data by means of a convolution code and restoring the code to a correct convolution code by means of the inverse interleave operation at the time of writing and reading. CONSTITUTION:A convolution coding means 1 encodes recording data once by means of the convolution code. An interleave means 2 makes an interleave control charging the time series order of codes at every block. After amplified by a write amplifier 3, it is written by a head H. At the time of reading, an inverse interleaving means 5 restores a regenerative signal to the convolution signal of the correct time series by means of the inverse interleave operation. By decoding an erroneous code with a Viterbi decoder or a sequential decoder, the magnetic disk where erroneous decision of Viterbi equalization hardly occurs is realized even if the S/N of the signal read from the magnetic disk is deteriorated.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a magnetic disk device, and more particularly to a magnetic disk device for decoding a reproduced signal from a magnetic head by using a maximum likelihood decoding method while performing error correction. In recent years, the S / N ratio of recording / reproducing signals has deteriorated and the intersymbol interference has increased with the downsizing and increasing the capacity of magnetic disk devices. In order to realize a highly reliable magnetic disk device under such severe conditions, an error correction technology for recording data is indispensable, and a magnetic disk device excellent in error correction for recording data is desired. ..

[0002]

2. Description of the Related Art Conventionally, intersymbol interference in magnetic recording and reproduction is regarded as a convolutional code, and demodulation is performed by a maximum likelihood decoding method, for example, a Viterbi decoding method. FIG. 6 shows the structure of a conventional magnetic recording / reproducing apparatus. Recorded data is RL
It is encoded by an L (Run Length Limited Code) encoder 81, recorded on a magnetic recording medium 64 by a write amplifier 82 and a head 63, read by a head 65 during reproduction, amplified by a read amplifier 66, and then Viterbi or the like. Demodulated data was obtained by the digitizer 67 and the RLL combiner 68. In the Viterbi decoding method, the inter-code interference of the read signal from the magnetic recording medium 64 is equalized to reduce the data error.

[0003]

However, in this Viterbi decoding method, the S / S of the read signal from the magnetic disk is
When N becomes worse, the number of erroneous determinations in Viterbi equalization increases, so that the error propagates to the contrary, and there is a problem that the error rate becomes worse than that before the equalization. Therefore, according to the present invention, the recorded data is once encoded with a convolutional code, and the write operation is performed by using an interleave operation for changing the time series order of the code for each block. The S / N of the read signal from the magnetic disk is restored by restoring the sequence convolutional code and performing error correction decoding by the Viterbi decoder or the sequential decoder.
It is an object of the present invention to provide a magnetic disk device in which erroneous determination in Viterbi equalization is unlikely to occur even when the power becomes worse.

[0004]

FIG. 1 shows the configuration of a magnetic disk device according to the present invention which achieves the above object. As shown in this figure, the present invention is a magnetic disk device which reproduces data recorded on a magnetic disk D by a magnetic head H and decodes a reproduced signal by a maximum likelihood decoding method. Convolutional encoding means 1 for converting into a convolutional code, and for the convolutionally encoded data,
Interleaving means 2 for performing an interleaving operation corresponding to the channels corresponding to the symbols of the convolutional code, a write amplifier 3 for amplifying the interleaved data and inputting it to the head H prepared for the channel, and the channel component. A read amplifier 4 for amplifying the prepared reproduced signal from the head H, and a deinterleave means 5 for performing a reverse interleave operation on the reproduced signal from the read amplifier 4 and returning the reproduced signal to convolutionally encoded reproduced data. And a maximum likelihood decoding means 6 for decoding the convolutionally encoded reproduction data while performing error correction by the maximum likelihood decoding method.

In the magnetic disk device, the recording data switching means 7 for time-dividing the recording data existing for the channels corresponding to the symbols of the convolutional code and transmitting the time-divided recording data to the interleaving means 2 for one channel, and the deinterleaving. There may further be provided multiplexer means 8 for returning the reproduction signal from the means 5 to the convolutional code for the channel in time division in synchronization with the operation of the recording data switching means 7. Further, the scrambling means 9 is provided in the preceding stage of the convolutional encoding means 1 and superimposes a random signal on the recorded data in the M sequence, and the scrambling means 9 is provided after the maximum likelihood decoding means 6 for the reproduced data. A descrambling means 10 for adding a random signal synchronized with the random signal in the scrambling means 9 to obtain an exclusive OR may be further provided.

[0006]

According to the magnetic disk device of the present invention, the error propagating at the time of Viterbi equalization is diffused by the operation of interleaving and approaches a random error. Then, strong error correction is performed in the decoding stage of the convolutional code, and by performing the interleave operation in this way, the defect of the convolutional code, which is weak against burst errors, is compensated, and the deterioration of the error correction capability is minimized. ..

[0007]

Embodiments of the present invention will be described in detail below with reference to the accompanying drawings. FIG. 2 is a block circuit diagram showing the configuration of an embodiment of the magnetic disk device of the present invention. First, the recording data writing system will be described. Reference numeral 20 denotes a convolutional encoder, which includes three-stage shift registers 21, 22, 23 and two exclusive OR circuits 24, 25. The convolutional encoder 20 is an encoder having a coding rate of 1/2 for outputting 2-bit data with respect to 1-bit data input. Reference numeral 30 is an interleave operation circuit, and shift registers 31, 3
2, 37, 38 and M × M series (4 × 4 series in FIG. 2)
Matrix circuits 33, 34, 35 and 36 are provided. Reference numerals 41 and 42 are write amplifiers, and 43 and 44 are write heads to the medium (disk) D.

The data read system includes a read circuit 50 having heads 51 and 52, read amplifiers 53 and 54, and Viterbi equalizers 55 and 56, and shift registers 71, 72, 77 and 7.
8 and M × M series matrix circuits 73, 74, 7
It is composed of an inverse interleave circuit 70 including 5,76 and a Viterbi decoder 80. In this embodiment, the heads 43 and 44 for the number of symbols of the convolutional code and the interleave circuit 30 are used to simultaneously perform recording and reproduction, and the transfer rate becomes high. In this example,
The symbols of the convolutional code obtained by the convolutional encoder 20 are M × M matrix, and the shift register 3
It is stored in the matrix circuits 33 and 34 via 1, 32, the matrix is transposed in the stored state, the order of the time series is exchanged, and stored in the other matrices 35 and 36. Then, after that, the matrices 35, 36
4 bits are read by the shift registers 37 and 38 from the write amplifiers 41 and 42 and the write heads 43 and 4, respectively.
4 is recorded on the disc D.

FIG. 3 illustrates the above-mentioned transposition by taking a 4 × 4 matrix circuit 33 as an example. In order to facilitate understanding, the input data are provided with numbers indicating the order. As shown in (a), the data is first stored in the shift register 31 bit by bit, and is transferred to the matrix circuit 33 when 4 bits are stored. The data stored in the matrix circuit 33 is shifted by 4 bits, and
When 6 bits of data are stored in the matrix circuit 31, they are transposed as shown in (b). This transposition is an operation of transposing the rows and columns of the data stored in the matrix circuit 31, and the transposed data is transferred to another matrix circuit 35. Then, the transposed data is read from the matrix circuit 35 to which the data has been transferred by the shift register 37 in units of 4 bits. The data read by the shift register 37 is the interleaved data. At this time, the following data is input from the shift register 33 to the original matrix circuit 31 by 4 bits at a time.

On the other hand, the signals read from the read heads 51 and 52 are amplified by the read amplifiers 53 and 54, and the intersymbol interference is removed by the Viterbi equalizers 55 and 56. The outputs of the Viterbi equalizers 55 and 56 are output to the shift registers 71 and 7, respectively.
The M × M matrix is stored in the matrix circuits 73 and 74 via 2, and the matrix is transposed in the stored state to change the order of the time series and stored in the other matrices 75 and 76. Then, after that, four bits are read from the matrices 75 and 76 by the shift registers 77 and 78 and returned to the original time series. Thereafter, the correct time-series convolutional codes output from the shift registers 77 and 78 are input to the Viterbi decoder 80 and decoded.

By the above operation, even if errors occur continuously due to the erroneous judgments of the Viterbi equalizers 55 and 56 during signal reproduction, the erroneous judgments are dispersed as errors of M intervals by the deinterleaving operation, and the Viterbi decoder is used. The code input to 80 is a convolutional code close to a random error. In this way
The weakness of the convolutional code against burst errors is compensated for, and deterioration of the error correction capability of the Viterbi decoder 80 is minimized.

FIG. 4 shows the structure of another embodiment of the magnetic disk device of the present invention. The same components as those of the magnetic disk device shown in FIG. 2 are designated by the same reference numerals. The magnetic disk device of this embodiment is different from the magnetic disk device shown in FIG. 2 in that the clock frequency is 2 in FIG.
That is, the number of circuit components is reduced by doubling. That is, the shift registers 32 and 38, the M × M series matrix circuits 34 and 36, the write amplifier 42, and the write head 44 which form the interleave operation circuit 30 of FIG. From the system, the read amplifier 52, the read head 54, the Viterbi equalizer 56, and the shift registers 72 and 7 forming the deinterleave operation circuit 70.
8 and M × M series matrix circuits 74 and 76
Has been deleted. And instead, the convolutional encoder 20 and the interleave operation circuit 3 in which the components are halved
A selector 26 for selecting the data of either the exclusive OR circuit 24 or 25 is provided between the Viterbi decoder 80 and the de-interleave operation circuit 70 'in which the components are halved. , A multiplexer 79 is inserted.

In this embodiment, the convolutional code is selected by the selector 26.
The basic operation is the same as that of the magnetic disk apparatus of the embodiment shown in FIG. 2 except that the read / write head, the interleave circuit, and the deinterleave circuit are reduced in size. Therefore, in this embodiment, transposition is performed for each M × M data block in the writing system, and the order of time series is exchanged. Further, in the reading system, the intersymbol interference of the signal read from the head 51 is removed by the Viterbi equalizer 55, the transposition is performed again for each M × M data block, and the original time series is restored. Further, it is returned to the convolutional code by the multiplexer 79 and is decoded by the Viterbi decoder 80.

FIG. 5 shows still another embodiment of the magnetic disk device of the present invention, in which functions are added to the magnetic disk device shown in FIG. Therefore, the same components as those in the embodiment of FIG. 4 are designated by the same reference numerals and the description thereof will be omitted. In the embodiment of FIG. 5, in order to spread the spectrum of the convolutional code and write it on the medium, an exclusive OR circuit 19 is provided in the preceding stage of the convolutional encoder 20, and the M-sequence (pseudo random) is added to this. The output of the pattern generator 18 is input. That is, in the embodiment of FIG. 5, the exclusive OR of the data and the output of the M-sequence generator 18 is taken by the exclusive OR circuit 19 to perform scrambling.

The read system is provided with an exclusive OR circuit 81 after the Viterbi decoder 80, to which the output of the M-sequence generator 82 is input. This M-sequence generator 82 is a write-system M-sequence generator 18
Which generates the same random number as the M sequence generator 18
It is designed to work in synchronization with. That is, in this embodiment, when the generated random numbers are exactly the same, the addition of the same random numbers through the exclusive OR circuit 81 cancels each other out to 0. Therefore, in the embodiment shown in FIG. 5, the exclusive OR of the demodulated data and the output of the same M-sequence (pseudo random pattern) generator 82 is obtained by the exclusive OR circuit 81 to output the original data. It has become.

When data is scrambled and recorded at the output of the M-sequence generator 18 as described above, the spectrum of the clock component appears in the read signal even if the recorded data is biased such that zeros continue. The clock can be surely reproduced. Therefore, in the embodiment of FIG. 5, the clock recovery circuit 90 is provided after the read amplifier 53.

As described above, according to the present invention, the recorded data is once encoded by the convolutional code, and the reproduction signal which is written and read out by the Viterbi equalization is read at the time of reading by using the interleave operation for changing the order of the time series of the code for each block. The S / N of the read signal from the magnetic disk is restored by restoring the convolutional code in the correct time series by the deinterleaving operation and performing error correction decoding by the Viterbi decoder or the sequential decoder.
Even if the value becomes worse, erroneous determination in Viterbi equalization is less likely to occur.

Further, according to the present invention, a large coding gain can be obtained by using a non-systematic code having a long constraint length as the convolutional code, and the recording density can be increased by the coding gain.

[0019]

As described above, according to the present invention,
Even if the S / N ratio of the read signal from the magnetic disk becomes poor, erroneous determination in Viterbi equalization is less likely to occur, and a large coding gain can be obtained by using a non-systematic code with a long constraint length as the convolutional code. Therefore, there is an effect that the recording density can be increased by the coding gain.

[Brief description of drawings]

FIG. 1 is a principle configuration diagram showing a configuration of a magnetic disk device of the present invention.

FIG. 2 is a block circuit diagram showing a configuration of an embodiment of a magnetic disk device of the present invention.

3A and 3B are explanatory views for explaining the transposition of FIG. 2;

FIG. 4 is a block circuit diagram showing the configuration of another embodiment of the magnetic disk device of the present invention.

FIG. 5 is a block circuit diagram showing the configuration of still another embodiment of the magnetic disk device of the present invention.

FIG. 6 is a block configuration diagram showing a configuration of a conventional magnetic recording / reproducing apparatus.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Convolutional encoding means 2 ... Interleaving means 3 ... Write amplifier 4 ... Read amplifier 5 ... Deinterleaving means 6 ... Maximum likelihood decoding means 7 ... Recording data switching means 8 ... Multiplexer means 9 ... Scramble means 10 ... Scramble means 18 , 82 ... M sequence generator 19, 24, 25, 81 ... Exclusive OR circuit 20 ... Convolutional encoder 21, 22, 23, 31, 32, 37, 38, 71, 7
2, 77, 78 ... Shift register 26 ... Selector 30, 30 '... Interleave operation circuit 33, 34, 35, 36, 73, 74, 75, 76 ... Matrix circuit 41, 42 ... Write amplifier 43, 44 ... Write Head 51, 52 ... Read head 53, 54 ... Read amplifier 55, 56 ... Viterbi equalizer 70, 70 '... Inverse interleave circuit 79 ... Multiplexer 80 ... Viterbi decoder 90 ... Clock recovery circuit

Claims (4)

[Claims] 1. A magnetic disk device which reproduces data recorded on a magnetic disk (D) by a magnetic head (H) and decodes a reproduced signal by a maximum likelihood decoding method, wherein the recording data is convoluted at the time of recording. Convolutional coding means (1) for converting into a code, interleaving means (2) for performing an interleaving operation corresponding to the channels corresponding to the symbols of the convolutional code on the convolutionally coded data, and interleaved operation. A write amplifier (3) for amplifying the obtained data and inputting to the head (H) prepared for the channel, and a read amplifier (4) for amplifying a reproduction signal from the head (H) prepared for the channel. And an inverse interleave operation for performing a reverse interleave operation on the reproduction signal from the reading amplifier (4) and returning the reproduction signal to convolutionally encoded reproduction data. (5) and, with respect to convolution coded reproduced data, the maximum likelihood decoding means for decoding while the error correction by the maximum likelihood decoding method (6)
A magnetic disk drive comprising: 2. Recording data switching means (7) for time-dividing recording data existing for channels corresponding to the symbols of the convolutional code and transmitting it to interleaving means (2) for one channel, and said deinterleaving means. Multiplexer unit (8) for returning the reproduced signal from (5) to the convolutional code for the channel in time division in synchronization with the operation of the recording data switching unit (7), further comprising: The magnetic disk device according to claim 1. 3. A scramble means (9), which is provided in the preceding stage of the convolutional coding means (1) and superimposes a random signal on the recording data with an M sequence, and a maximum likelihood decoding means (6). A descrambling means (10) which is provided in a subsequent stage, and which adds a random signal synchronized with the random signal in the scrambling means (9) to the reproduction data to take an exclusive OR, and The magnetic disk device according to claim 1. 4. The magnetic disk device according to claim 3, further comprising a clock reproducing means (11) which is provided after the read amplifier (4) and reproduces a clock signal from a reproduced signal.