JPH10106157A - Digital recording and reproducing method and digital recording and reproducing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To secure the reliability under a condition in which resolution is low by removing the error propagation of a DFE(decision feedback equalization). SOLUTION: This device is provided with equalization means 9, 14 in which feedback system are not present and which obtain an identification result A by an equalization characteristic, an equalization means 15 obtaining the identification result B of the DFE, a means 17-T obtaining a corrected identification result C of the identification A based on the identification result B, comparator means 16, 20 comparing the identification results A and B and comparator means 18, 19 comparing the identification results B and C. A switcher 22 to be controlled by a controller 21 outputs the identification result B of the DFE normally and when the error propagation of the DFE is predicted by the comparision results of comparing means 16, 20; 18, 19 it is changed over to another identification result A to output the result A.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a digital recording / reproducing method and a digital recording / reproducing apparatus, and more particularly, to an improvement of a decision feedback equalizer.

[0002]

2. Description of the Related Art As the operation speed of computers increases and application software that requires a large-capacity memory is used, the demand for higher recording density of small magnetic disks is increasing. . Because of this,
As a signal processing method of a recording / reproducing system, a PRML method in which a partial response class 4 (hereinafter, referred to as PR4) equalization method and maximum likelihood decoding are combined has attracted attention, and the S / N is about 3 dB compared to a conventional peak detection method. Improvements have been realized. Regarding these technical features, for example, Journal of the Japan Society of Applied Magnetics, Vol. 18, No. 4, (1994) 85
7-862, "PRML and Coding Techniques".

On the other hand, a decision feedback equalizer (Decision feedbac
Conventionally, k equalization (hereinafter referred to as DFE) has been used in a digital signal receiving system in a communication cable, but has recently attracted attention for signal processing in a recording and reproducing system in the magnetic disk. For example, Journal
of Applied Physics, Vol. 79, No. 8, April 1996
"Signal processin" by K. Fisher and C. Modlin
g for 10Gb / in magnetic disk recording and beyond "
The DFE shows the superiority of DFE,
It is reported that the area density can be improved 1.1 times to 1.2 times as compared with RML.

[0004]

The DFE is a method of calculating the amount of interference expected in a transmission system on the basis of the identification result, and extracting a signal component by subtracting it from an incoming transmission signal. Therefore, when the identification result is wrong for some reason, the wrong interference is subtracted, and the error is enlarged. This spread of errors is called error propagation.

In a communication system, the ratio of redundancy of an error correction code included in a signal sequence to be transmitted is high, and the error rate before correction is 1
0 in about ^-4 also has entered the correction range, not to much of a problem the error DenHata. However, in a recording / reproducing system for a magnetic disk that requires a small bit error rate, error propagation of DFE becomes a serious problem. An object of the present invention is to provide a configuration of a DFE that can be applied to a recording / reproducing system of a magnetic disk that requires a small bit error rate while coping with the error propagation of the DFE, and further has a large capacity using the DFE. An object of the present invention is to provide a magnetic disk drive.

[0006]

According to the present invention, first equalizing means for obtaining an identification result A by an equalization characteristic EQ1 without a feedback system such as PR4, and identification by an equalization characteristic with a feedback system such as a DFE. A second equalizer for obtaining a result B; and a third equalizer for obtaining a discrimination result C by the equalization characteristic EQ2 capable of coping with a higher recording density than the equalization characteristic of the first equalizer. The types A, B, and C are compared. And usually, the identification result B by DEF is used,
When the error propagation of the DFE is expected by comparing the identification results A, B, and C, the digital recording recorded at a high recording density is highly reliable by switching to another identification result and removing the error propagation of the DFE. It is possible to play with the nature.

That is, according to the present invention, in a digital recording / reproducing method for reproducing a digital recording recorded on a recording medium and decoding an original recording signal, an identification result A based on a first equalization characteristic without a feedback system is provided. An identification result B based on the second equalization characteristic having a feedback system and an identification result C based on the third equalization characteristic capable of coping with a higher recording density than the first equalization characteristic are obtained, and the identification results A, B, and C are obtained. And selecting an identification result having few code errors as a final identification result.

The selection of the identification results A, B, C
When it is determined that error propagation has occurred in the identification result B by the comparison between the identification result B and the identification result B and the identification result B, the identification result A is selected as the final identification result. The identification result B or the identification result C can be selected as the final identification result.

Further, a difference between the identification results A and B and a difference between the identification results B and C are obtained, and when the difference between the identification results A and B is smaller than a set value, the identification result B is selected as a final identification result. If the difference between the identification results A and B is larger than the set value and the difference between the identification results B and C is smaller than the set value, the identification result B or C is selected as the final identification result.
If the difference between B and C is greater than the set value and the difference between identification results B and C is greater than the set value, identification result A can be selected as the final identification result.

The first equalization characteristic can be a PR4 equalization characteristic, and the third equalization characteristic can be an extended PR4 equalization characteristic. The pre-filter characteristic of the second equalization characteristic is
Using D as a delay operator, it can be 1 + D−D ³ −D ⁴ .

Further, according to the present invention, there is provided a digital recording / reproducing apparatus for reproducing a digital recording recorded on a recording medium and decoding an original recording signal to obtain an identification result A by an equalization characteristic without a feedback system. Equalizing means, second equalizing means for obtaining an identification result B based on an equalizing characteristic having a feedback system, and identification result C using an equalizing characteristic capable of coping with a higher recording density than the equalizing characteristic of the first equalizing means Third to get
, A first comparing means for comparing the identification results A and B, a second comparing means for comparing the identification results B and C, and a first comparing means and a second comparing means. Selecting means for selecting a final identification result from three types of identification results A, B, and C based on the comparison result. The third equalization means can obtain an identification result C obtained by modifying the identification result A by processing the output of the first equalization means.

The first comparing means has means for counting the number of mismatches between the identification results A and B, and the second comparing means has means for counting the number of mismatches between the identification results B and C. Can be. The selecting means selects the identification result B when the count value of the first comparing means is smaller than the set value,
When the count value of the first comparing means is larger than the set value and the count value of the second comparing means is smaller than the set value, the identification result B or C is selected, and the count value of the first comparing means is larger than the set value. When the count value of the second comparing means is larger than the set value, the identification result A can be selected. The main part of the digital recording / reproducing apparatus according to the present invention can be realized on a semiconductor integrated circuit.

[0013]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a signal system diagram when the present invention is applied to a magnetic disk recording / reproducing apparatus. Data from a host computer (not shown) is input to a magnetic disk device, and is converted by an encoder 1 into a recording code suitable for a magnetic disk recording medium system. The output of the encoder 1 is recorded on a recording medium 4 via an amplifier circuit 2 and a recording head 3. At the time of reproduction, a signal is read from the recording medium 4 via the reproduction head 5 and the amplifier circuit 6, and noise other than the signal band is removed by the analog filter 7. Also, the filter 7 is designed to prevent the frequency characteristic in the signal band from deteriorating due to the physical separation between the recording medium 4 and the reproducing head 5.
The compensation is performed. The output of the analog filter 7 is output from the analog / digital converter 8 at a predetermined speed.
It is sampled, quantized, and converted to, for example, a 6-bit digital value. Subsequent signal processing is digital processing and is less affected by ambient temperature and external noise.

The obtained digital value is converted by the PR4 equalizing digital filter 9 so that the transmission characteristic of the recording / reproducing system is (1).
−D ² ) The coefficients of each stage of the filter are set so as to be −D ² ). Here, D is a delay operator. (1-D ² ) is a characteristic of a so-called partial response class 4, and a predetermined reproduction clock synchronized with the signal read by the reproduction clock generation circuit 10 can be extracted from the filter output. This clock is supplied to the A / D converter 8, the PR4 equalizing digital filter 9, and the DFE circuit 1.
5 and the like are supplied to a reproduction system circuit system. Since the digital filter 9 provides a PR4 equalized output, the maximum likelihood detecting circuit 14 having two states generates an identification result A in which the influence of noise superimposed on the equalized output is minimized. Is obtained.

On the other hand, a DFE circuit 1 suitable for high-density recording
5 is also supplied with the output of the A / D converter 8 at the same time.
Is obtained. Generally, DFE is PR4
Compared with the PRML identification output A that combines equalization and maximum likelihood decoding, an identification result B with less code error is provided in high-density recording. However, once a code error occurs, the result is fed back, so that the error propagates. Therefore, in the normal state, DFE
It is an advantageous method to use the identification result B of the above as the final identification result FD and the PR4ML output A as the final identification result FD when it is considered that error propagation has occurred.

In order to determine whether or not error propagation has occurred, first, the identification result A of PR4ML and the identification result B of DFE are compared to find a place where the identification results A and B have a large difference. The cause of the big difference is (1) DF
E has an error propagation, and (2) the identification result A has many code errors because the margin of the recording / reproducing system is small for PR4 equalization. In order to determine whether the cause of the large difference is the above (2), the equalized output of the equalization characteristic EQ2 which can cope with the higher density recording than the PR4 equalization is calculated, and the maximum likelihood detection (ML) is performed.
Then, the identification result C ′ is obtained, and the difference from the identification result B is observed. If the difference between the identification results C ′ and B is small, the above condition (2) is satisfied.
Can be determined. If the difference is large, the condition (1) is estimated, and the PR4ML identification result is adopted.

However, if the processing (EQ2 + ML) of calculating the equalized output of the equalization characteristic EQ2 and performing the maximum likelihood detection (ML) as it is to obtain the identification result C 'increases the power consumption of the circuit, Here, the equalization characteristic EQ2 is calculated based on the DFE identification result B, and the output EQ
The difference between 2 ′ and the equalized output EQ2 from PR4 is obtained, and maximum likelihood detection is performed on a code error (ER) corresponding to the difference. The identification result A is corrected by this detection result, and (E
An identification result C corresponding to (Q2 + ML) processing is obtained. In the maximum likelihood detection for a code error, the data to be handled is almost a value close to 0, and there is an advantage that the power consumption of the maximum likelihood circuit can be reduced. In this example, the equalization characteristic EQ without a feedback system
PR4 equalization as 1: (1-D ² ), and extended PR4 as equalization characteristics EQ2 corresponding to high density: (1-D ² ) (1+
D) is used.

The circuit system 17-T is a circuit for obtaining the identification result C. First, processing (1 + D) is performed on the PR4 equalized output by the arithmetic circuit 11. As a result, (1-
D ² ) (1 + D) output is obtained from the arithmetic circuit 11. On the other hand, from the identification result B of the DFE 15, similarly, (1-D ² )
The (1 + D) equalized waveform output is generated by the waveform generation circuit 18. The waveform generation circuit 18 includes a 4-bit shift register 18-1 and a memory 18-2 as shown in FIG. The identification result B is input to the shift register 18-1. As shown in FIG. 3, (1) for the input sequence B
The operation result S of −D ² ) (1 + D) can be calculated in advance. (1-
D ² ) (1 + D) equalized waveform output is 1,1, −1, −1
It is. The memory output becomes the equalized output S for 4 bits of the input sequence B.

Since the equalized output S is based on the result of DFE discrimination, it can be considered as a correct equalized output with few code errors in normal times. On the other hand, PR4ML and (1
+ D) The calculated equalized output contains noise. When these are subtracted by the subtraction circuit 12, noise including a code error is output. When a code error is detected by the maximum likelihood detecting circuit 13 and the modulo 2 is added to the PR4ML identification result A by the exclusive OR circuit 17, PR4M
An identification result C obtained by modifying the result of L with the result of DFE is obtained. However, since the result C is calculated assuming that the identification result of the DFE is correct, if an error propagation occurs in the DFE, a code error is added conversely.

In FIG. 1, delay circuits 24-1 and 24-24 are used.
Reference numerals 2 and 24-3 are for adjusting the timing for calculating each identification result with another identification result. The exclusive OR circuit 16 compares the PR4ML identification result A with the DFE identification result B, and the two identification results have different output bits. Similarly, the identification result B and the identification result C are also compared by the exclusive OR circuit 18. The number of different bits is counted by the counters 19 and 20 in codeword units. The respective counting results are input to the controller 21 at the same timing by the delay circuit 24-4, the final identification result FD is obtained via the switcher 22, and the identification result FD is converted to the original information data by the decoder 23.

Main part 1 from encoder 1 to decoder 23
-23-T can be incorporated into one semiconductor integrated circuit and commercialized as an integrated circuit for signal processing for a recording / reproducing system. The circuit system 22-T including the controller 21 is a means for selecting an identification result. Since the setting method of the controller 21 is the core technology of the present invention, it will be described in detail below. Each of the counter 19 and the counter 20 is provided with a predetermined set value, and the controller 21 controls the switcher 19 depending on whether the count value is larger or smaller than the set value. FIG. 4 shows the control method.

The fact that the value of the counter 20 is smaller than the set value means that there is little difference between the identification outputs A and B,
Almost no code error occurs in DFE15, PR4ML
It is presumed that the code error of DFE has been counted, so that the DFE is
Select output B.

On the other hand, if the value of the counter 20 is larger than the set value, it is determined whether error propagation has occurred in the DFE 15 or not.
Either there are many errors in R4ML. To distinguish this, the value of the counter 19 is referred to. DFE when the value of the counter 19 is larger than the set value
15, it is predicted that error propagation has occurred, and PR
The identification result A of 4ML is adopted. If the value of the counter 19 is smaller than the set value, the original PR4ML has many code errors due to the influence of noise, which means that the corrected PR4ML output and the DFE output match well.
The correction result C is defined as a final identification result FD. At this time, D
An equivalent result can be obtained by using the identification result B of the FE 15 as the final identification result FD.

FIG. 5 shows the basic configuration of the DFE 15. When a response waveform of an isolated pulse is input, the pre-filter 15-1 suppresses noise emphasis as much as possible,
Waveform shaping is performed, and filter coefficients are set so that waveform interference appears after the bit. That is, the response of the isolated pulse is 1 + k ₁ D + k ₂ D ² + k ₃ D ³ + K ₄ D ⁴ + ‥
Given by ‥. Here, k _i is a predetermined coefficient, and i is a finite integer. The recording code is _ai = 0 or 1,
a ₀ + a ₁ D + a ₂ D ² + a ₃ D ³ + ‥‥. The multiplication of these two equations is the output of the pre-filter 15-1. As will be described later, the signal from which the waveform interference has been removed is sent to the identification circuit 15-3. When the identification result B is obtained by the identification circuit 15-3, a predetermined number of bits of the B sequence is input to the shift register 15-4, and the waveform interference component k ₁ D + k ₂ D ² + k ₃ D ³ + K _{4 with} respect to the set isolated pulse. D ⁴ +
The waveform interference for the predetermined bits of the B sequence is calculated in advance based on ‥‥, and when the predetermined number of bits of the B sequence is given as an address to the memory 15-5, the waveform interference is output and the subtraction circuit 15- 2, an output without waveform interference is obtained.

Since the DFE estimates the waveform interference based on the identification result, if the identification result is erroneous, an erroneous waveform interference is output, and the error may be enlarged. Therefore, there has been a scheme to reduce the amount of waveform interference to be fed back and to introduce an idea of maximum likelihood decoding into the discriminating circuit to reduce errors in the discrimination result as much as possible. The present invention is applicable to such a DFE. In that case, it goes without saying that the timing adjustment between the DFE identification result and the other identification results must be changed from the example here.

FIG. 6 is a circuit diagram of another embodiment of the present invention. The circuit of FIG. 6 has three identification results A,
The circuit configuration for obtaining B and C is shown, and the other parts are the same as those in FIG. Also, circuits having the same numbers as those in FIG. 1 have the same functions as those in FIG. A / D
The output of the converter 8 is converted by the digital filter 9 into PR4.
Undergo equalization. The feature of this embodiment is that the DFE 15 '
Circuit for setting the response characteristic of the pre-filter to 1 + D−D ³ −D ⁴ and obtaining an equalized waveform output from the identification output B of the DFE 15 ′ in order to obtain the identification result C based on the identification result of the DFE 15 ′ In the circuit system of DFE15 ',
The point is to reduce the circuit scale. Adder circuit 15-1-
3 is the output of PR4 together with the delay elements 15-1-1 and 1-1-1.
The output of 5-1-2 is added and (1-D ² ) (1 + D +
D ²⁾ = 1 + gives an equalized output of the D-D ³ -D ^4. Subtraction circuit 15-2 and identification circuit 15 forming a feedback system
-3, the shift register 15-4, and the memory 15-5, the interference correction amount DD- ³ based on the DFE identification result B.
The output corresponding to -D ⁴ is obtained, with the B output corresponding to the D ² from the shift register 15-4, the adding circuit 1
Add in 8-1.

As a result, the DFE of D + D ² -D ³ -D ⁴
An equalized output based on output B is obtained. Similarly, P
D + D ² − from the R4 equalized output by the adder circuit 11-1
An equalized output of D ³ -D ⁴ is obtained. These are input to the subtraction circuit 12 at the same timing via the timing adjustment delay element 11-2. A signal which seems to be a code error generated due to a difference between the two equalized outputs is detected by the maximum likelihood detecting circuit 13 and a logical circuit 17 obtains a corrected identification result C by taking an exclusive OR with the identification result A of PR4. . The method of selecting the three identification results A, B, and C is the same as in the first embodiment.

A repetition signal of a square wave signal is recorded on a magnetic disk, and the output amplitude of the magnetic disk when the repetition frequency is sufficiently low is 100%, and the output when the repetition frequency is the highest in actual use is 15% or less. In the magnetic disk device, the calculation result that the DFE has fewer code errors than PR4ML is obtained. The present invention provides a DFE while utilizing the capability of the DFE at such a low resolution.
PR4ML when an error transmission, which is the problem of
To provide a highly reliable magnetic disk drive. Note that even if an error propagation occurs when a random signal is being reproduced, the propagation is short. Rather, there is no way to prevent error propagation when the same pattern is repeated, which impairs reliability. vice versa,
The present invention is effective because PR4 and the like have the same pattern with little waveform distortion and the like, so that noise can be easily removed and a highly reliable identification result can be obtained.

Since the maximum likelihood detection circuit 13 detects a code error, data to be handled is almost zero.
The power required to store data in a memory (not shown) required for maximum likelihood detection is small, and the maximum likelihood decoding circuit with a large number of states consumes little power. It is within the scope of the present invention to use the concept of the present invention even when the circuits are parallelized in order to handle high-speed transfer data. In the above embodiment, PR4
, But EPR4 (1 + D−D ² −D ³ ) equalization is used as a basic equalization, and 1 + D−D is used as an equalization for improving a code error.
³ -D ⁴ using the method are also within the scope of the present invention. The present invention is also applicable to various PR system modifications. In the description here, the counting range of the counters 19 and 20 is a codeword unit. However, since the setting of this counting range depends on the assumed code error rate of the recording / reproducing system, it is set to a plurality of codewords. In some cases.

[0030]

According to the present invention, the reliability of a magnetic disk device having a low resolution can be ensured. In other words, the spacing between the magnetic disk and the head, which is one of the parameters giving the resolution, can be given a margin, which has the effect of preventing an accident such as a head crash.

[Brief description of the drawings]

FIG. 1 is a signal system diagram of an example of a recording / reproducing system according to the present invention.

FIG. 2 is a circuit configuration diagram for generating an equalized waveform added to improve a code error.

FIG. 3 is a diagram showing a process of calculating an equalized waveform added to improve a code error.

FIG. 4 is a diagram showing operating conditions of a controller.

FIG. 5 is a circuit configuration diagram of a DFE.

FIG. 6 is a signal system diagram of another example of the recording / reproducing system according to the present invention.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Encoder, 3 ... Magnetic recording head, 4 ... Recording medium, 5
... reproduction head, 8 ... A / D converter, 9 ... PR4 equalizer,
14 maximum likelihood decoder, 15 DFE circuit, 18 equalized waveform generation circuit, 19, 20 counter, 21 controller, 22 switcher, 23 decoder

Claims (13)

[Claims] 1. A digital recording / reproducing method for reproducing digital recording recorded on a recording medium and decoding an original recording signal, comprising: an identification result A based on a first equalization characteristic without a feedback system; and a feedback system. The identification result B based on the second equalization characteristic and the identification result C based on the third equalization characteristic that can cope with a higher recording density than the first equalization characteristic are obtained, and the identification results A, B, and C are compared. A digital recording / reproducing method characterized by selecting an identification result having few code errors as a final identification result. 2. When a comparison between the identification results A and B and a comparison between the identification results B and C determine that error propagation has occurred in the identification result B, the identification result A
2. The digital recording / reproducing method according to claim 1, wherein the identification result B or the identification result C is selected as a final identification result at any other time. 3. A difference between the identification results A and B and a difference between the identification results B and C are determined, and when the difference between the identification results A and B is smaller than a set value, the identification result B is finalized. When the difference between the identification results A and B is greater than a set value and the difference between the identification results B and C is smaller than a set value, the identification result B or C is selected as a final identification result. 2. The identification result A is selected as a final identification result when a difference between the identification results A and B is larger than a set value and a difference between the identification results B and C is larger than a set value. Digital recording and reproduction method. 4. The digital recording / reproducing method according to claim 1, wherein said first equalization characteristic is a PR4 equalization characteristic. 5. The digital recording / reproducing method according to claim 1, wherein said third equalization characteristic is an extended PR4 equalization characteristic. 6. The pre-filter characteristic of the second equalization characteristic, wherein D is a delay operator, is 1 + D−D ³ −D ^4. The digital recording / reproducing method as described. 7. A digital recording / reproducing apparatus for reproducing digital recording recorded on a recording medium and decoding an original recording signal, comprising: a first equalizing means for obtaining an identification result A by an equalizing characteristic without a feedback system; A second equalizing means for obtaining an identification result B by an equalization characteristic having a feedback system, and a second equalization means for obtaining an identification result C by an equalization characteristic capable of coping with a higher recording density than the equalization characteristic of the first equalization means. 3, the first comparing means for comparing the identification results A and B, the second comparing means for comparing the identification results B and C, the first comparing means and the second A selecting means for selecting a final identification result from three types of identification results A, B, and C based on the comparison result of the comparison means. 8. The digital processing apparatus according to claim 7, wherein said third equalizing means obtains an identification result C obtained by modifying an identification result A by processing an output of said first equalizing means. Recording and playback device. 9. The first comparing means has means for counting the number of mismatches between the identification results A and B, and the second comparing means counts the number of mismatches between the identification results B and C. 9. The digital recording / reproducing apparatus according to claim 7, further comprising means for counting. 10. The selection means selects the identification result B when the count value of the first comparison means is smaller than a set value, and the count value of the second comparison means is larger than the set value. When the count value of the comparing means is smaller than the set value, the identification result B or C is selected, and the count value of the first comparing means is larger than the set value and the count value of the second comparing means is larger than the set value. 10. The digital recording / reproducing apparatus according to claim 9, wherein when the value is larger, the identification result A is selected. 11. The apparatus according to claim 7, wherein said first equalizer performs PR4 equalization, and said third equalizer performs extended PR4 equalization. Digital recording and playback device. 12. A digital recording apparatus using the digital recording / reproducing method according to claim 1. 13. A semiconductor integrated circuit using the digital recording / reproducing method according to claim 1. Description: