JPH0489664A - Method for converting and detecting data - Google Patents

Abstract

PURPOSE:To reduce an error rate in the case of reproduction and to improve recording density by selecting a code, for which the total sum (modulation code distance) of the absolute values of difference mutually between the codes of intermediate sequences is more than the prescribed-fold standard value of weighting, as a code in the preceding part at least of a converted data. CONSTITUTION:An N number of codes in the converted data are divided into an Ns number of codes in a preceding part and an Nt number of codes in a following part, and among the Ns number of codes in the preceding part or the Nt number of codes in the following part, an (n) number of continuous codes, which number is the allowable number of interference between codes in a transmission line, are weighted with distribution to be linearly decreased from the center. When the (n) number of weighted continuous codes are successively added and the intermediate sequence is formed, as the code in the preceding part of the converted data at least, the code, for which the total sum of the absolute values of difference mutually between the codes of the intermediate sequences is more than the prescribed fold standard value of weighting, is selected. Thus, the pattern distance between code patterns can be enlarged, the error rate in reproduction can be reduced while using an active medium and a recording / reproducing device, and the recording density can be improved.

Description

[Detailed description of the invention] The invention will be explained in the following order.

A. Industrial field of application B0 Overview of the invention C1 Prior art 1 Problems to be solved by the invention 81 Means for solving the problems F1 Effect G. Example G1 Structure of the example Figure 3) G2 Operation of one embodiment (Figures 1 to 9) C3 Configuration of another embodiment (1st O
Figure) C4 Operation of Other Embodiments (Figures 10 to 13) H0 Effects of the Invention A, Field of Industrial Application This invention relates to a data conversion and detection method suitable for high-density recording.

B1 Summary of the Invention The first invention provides a data conversion method for converting original data in units of M bits to be recorded on a medium into converted data in units of N (>M) bits. The center is divided into Ns leading parts and Nt trailing parts, and among the Ns leading parts or Nt trailing parts, the center When an intermediate sequence is formed by sequentially adding n consecutive weighted codes using a distribution that linearly decreases from By selecting the sum of values (modulation code distance) that is a predetermined times or more of the weighting standard value,
By increasing the pattern distance between code patterns, the error rate during reproduction can be reduced and the recording density can be significantly improved while using current media and recording/reproducing devices.

A second aspect of the present invention provides a data detection method for detecting converted data from a reproduced signal of a medium in which converted data in units of N bits is recorded, in which transmission of a reproduction system of the medium is performed so as to allow n intersymbol interferences. In addition to setting the characteristics, the data to be detected is divided into N bit units of converted data, in units of Ns bits in the leading part and in units of Nt bits in the succeeding part,
When detecting at least the leading part of the converted data, the pattern of the reproduced signal corresponding to the data to be detected is compared with the code pattern group corresponding to each converted data, and an approximate pattern is selected. This makes it possible to reliably detect converted data recorded at an extremely high density with a low error rate while using a recording/reproducing device.

C8 Prior Art Conventionally, in digital magnetic recording, the binary values 1"1" and "0" of a digital signal are basically associated with the polarity of magnetization or the presence or absence of magnetization reversal. Then, depending on the physical characteristics of the recording medium, the transmission band of the system, etc., the data is encoded according to various modulation (writing) methods such as NRZ FM MFMGCR (8-10 conversion).
The binary value and magnetization are made to correspond appropriately on a bit-by-bit basis.

In general, these modulation methods require a large minimum magnetization reversal interval (Tmin) and a large maximum magnetization reversal interval (Tmin).
max) is small, local recording density changes are small, direct current components are small, and encoding conversion efficiency is high.

Further, during demodulation (reading), differential detection or integral detection is used depending on the amount of DC component of the modulation code, and data is demodulated by detection in units of 1 bit.

Conventional digital magnetic recording as described above is based on the assumption that there is no intersymbol interference, and requires that the level of the reproduced signal in the high frequency range be sufficiently high.

In other words, the highest repetition frequency of the recording signal and the density of recorded information are determined by the S/H of the high frequency reproduction signal corresponding to Tm1n of various modulation codes.

For this reason, as shown in FIG. 14, the current maximum repetition frequency f max is set at a portion where a high S/N can be obtained in a decreasing region of the reproduction level-frequency characteristic. In this descending region, the level of the reproduced signal decreases at a gradient of, for example, 126B10ct due to various losses during recording and reproduction.

Further, an ideal transmission characteristic (frequency spectrum) as shown in FIG. 15A is required, and a sinusoidal descending equalization characteristic satisfying Nyquist's first criterion as shown in FIG. 15A is used.

In addition, in the same figure, the frequency fo is Tm of the modulation code
This corresponds to 1n.

Further, Nyquist's first criterion is a condition in which when a waveform is sampled at regular intervals on the receiving side, sampled values other than the center become 0.

For example, the current digital audio tape recorder (D
In the case of 8-10 conversion, which is used in AT), the encoding conversion efficiency is [8/10:], and the relative speed between the coated metal (MP) tape and the magnetic head is over 3 m/s, The maximum repetition frequency f max was set to 4.7 MHz, the gap length was 0.25 μm, and the recording wavelength was 0.6
It becomes 7 μm. In this case, the track pitch is 10 to 15
In μm, the linear recording density is generally limited to about 60 to 3 Q kbpi.

D1 Problem to be solved by the invention By the way, when recording at high density, the maximum repetition frequency f
It is possible to increase max.

However, if you try to double the recording density, for example,
At a repetition frequency of 2fmax, as is clear from FIG. 14 above, there is a problem in that the level of the reproduced signal decreases, the S/N ratio deteriorates significantly, and data detection becomes impossible.

As mentioned above, current magnetic systems use recording media and conversion devices at their limits, so it is possible to reduce various losses during recording and playback and significantly improve the level of playback signals in high frequencies. It is extremely difficult to do so.

On the other hand, when increasing the recording density, the following problem of intersymbol interference in the reproduced waveform occurs.

When one magnetization reversal exists in isolation on a magnetic medium,
As a reproduced signal, a pulse-like voltage waveform (isolated pulse) as shown in FIG. 16 is obtained. The waveform of this isolated pulse, that is, the impulse response, is approximated to a Lawrence type waveform as shown in the following equation (1), and its spread on the time axis (pulse width) depends on the recording system/playback system used. It is determined by the overall transmission characteristics of the waveform and the magnetic medium, and is usually expressed as the half-width Wh1 at a level of 50% of the peak value or the width wb at a base level of substantially 0%.

f(t) = 1/ (1-'-, (t/ to)2)
” = (1) When multiple magnetization reversals exist consecutively at regular intervals, if the recording density is coarse, there will be no interference between adjacent pulses during reproduction, and the reproduced signal will consist of alternating isolated pulses as described above. It is just a series of inverted images.

The recording density has become considerably high, as shown in Figure 17.
Pulse width wb when the interval between adjacent pulses is at the base level
When the width is narrowed to 172, the tails of adjacent pulses overlap, and the waveform of the reproduced signal becomes quite different from that of an isolated pulse.

As is clear from the figure, at this stage, information on the peak value of each pulse is preserved without any loss, and although there is interference between waveforms, no intersymbol interference occurs.

When the recording density becomes higher than the state shown in FIG. 17, the peak value of the reproduced signal decreases, and nonlinear intersymbol interference (peak shift) occurs in which the interval between peak positions increases.

As the recording density further increases, for example, as shown in Figure 18, when the interval between adjacent pulses narrows to 174 degrees of the pulse width wb at the base level, the reproduced pulse waveform becomes close to a sine wave, and the peak value is also significantly reduced. In this case as well, intersymbol interference occurs that destroys information on the peak value of each pulse.

By the way, as a method that utilizes intersymbol interference, a partial response (Partial Response) method% type% is used.In this method, for example, the first
As shown in FIG. 9, there is an advantage that the frequency spectrum is limited to the Nyquist bandwidth and high frequency components are not required.

The characteristics shown in FIG. 19 correspond to partial response class 4 (modified duobinary) and are expressed as in the following equation (2).

Pr(1,0,-1) = sin(2yr f
/fo) = - (2) However, since the various modulation codes mentioned above do not themselves assume intersymbol interference, even if the partial response method is applied, its advantages cannot be fully utilized. There was a problem.

Further, in some types of modulation codes, Tmax becomes infinite, and even if a partial response method is applied, there is a problem that functions necessary for system configuration such as override or clock regeneration cannot be realized.

In view of this, an object of the present invention is to provide a data conversion method that can reduce the error rate during reproduction and improve the recording density while using current recording media and recording/reproducing devices. and a detection method.

Means for Solving Problem E1 The first invention provides a data conversion method for converting original data in units of M bits to be recorded on a medium into converted data in units of N (>M) bits. The number of inter-symbol interferences to be allowed is calculated according to the characteristics of the transmission path. Weighting is performed on the equal n consecutive codes so as to decrease linearly from the standard value at the center of the distribution, and the weighted n consecutive codes are sequentially added to obtain Ns - n + 1 bit unit or When an intermediate sequence of Nt-n+1 focus units is formed, the code of at least the preceding part of the conversion data is such that the sum of the absolute values of the differences between the codes between the intermediate sequences is at least a predetermined times the weighting standard value. This is the data conversion method selected.

A second aspect of the present invention provides a data detection method for detecting converted data from a reproduced signal of a medium in which converted data in units of N bits is recorded, in which transmission of a reproduction system of the medium is performed so as to allow n intersymbol interferences. In addition to setting the characteristics, the data to be detected is divided into N bit units of converted data, in units of Ns bits in the leading part and in units of Nt bits in the succeeding part,
When detecting at least the leading part of the converted data, the signal pattern of the reproduced signal corresponding to the data to be detected is compared with a plurality of predetermined patterns limited by the conversion data, and the signal pattern is approximated from the plurality of predetermined patterns. This is a data detection method in which data is detected by selecting an approximate pattern.

F1 Effect According to the present invention, the error rate during reproduction is reduced and the recording density is significantly improved while using current recording media and recording/reproducing devices.

G. Embodiment Hereinafter, an embodiment of the data conversion and detection method according to the present invention will be described with reference to FIGS. 1 to 9.

Configuration of G1-Embodiment The overall configuration of a magnetic system to which an embodiment of the present invention is applied is shown in FIG. 1, and the configuration of its main parts is shown in FIGS. 2 and 3.
As shown in the figure.

In FIG. 1, (10) is a recording system, and the terminal IN
Analog signals such as audio signals and video signals from the
2), and recording data conforming to the system format is generated. The output of the generation circuit (12) is supplied to the data conversion circuit (13), and the symbol data output from this conversion circuit (13) is supplied to the magnetic node (1) via the recording amplifier (14). and magnetic tape MT
recorded directly.

As shown in FIG. 2, the data conversion circuit (13) includes a pair of ROM tables (13a) and (13b),
A pair of conversion codes having different DC component distributions as shown in Tables 1 and 2 below are stored respectively. The conversion code (symbol data) read from the ROM table (13a), (13b) is sent to the DC component adjustment circuit (13b).
3c) is selectively derived via a switch (13S) controlled by .

(20) is a reproduction system, in which the signal (differential waveform) reproduced from the magnetic tape MT by the magnetic head (2) is processed through a reproduction amplifier (21) and subjected to waveform equalization consisting of an integrator and a low-pass filter. It is supplied to the circuit (22). This equalization circuit (
22) is supplied to the A-D converter (23), where it is converted into, for example, 8-bit data, and is also supplied to the PLL circuit (24). The output of the PLL circuit (24) is supplied as a signal.

(30) is a data detection circuit, the detailed configuration of which will be described later. The output of the A, -D converter (23) is supplied to this data detection circuit (30), and the detected symbol data is supplied to the decoding circuit (24), where it is decoded into original data and sent to the output terminal 011T. derived.

In FIG. 3, a series of pattern data is commonly supplied to the synchronization detection circuit (31) and the symbol division circuit 32) from the input terminal (30i) of the data detection circuit (30A). This division circuit (32) is controlled by the detection output of the synchronization detection circuit (31) and is
For each pattern data corresponding to each conversion data (symbol data) as shown in the table, the preceding part Ss and the succeeding part St
and the leading part Ss is divided into a plurality of subtractors (
33a) to (33s).

(34a) to (34s) are reference value ROMs in which waveform values of each preceding part code pattern selected as a reference are stored, respectively. The output of the synchronization detection circuit (31) is commonly supplied to each ROM (34a) to (34s),
The outputs of the ROMs (34a) to (34s) are supplied to corresponding subtracters (33a) to (33s), respectively.

(35a) to (35s) are pattern distance calculation circuits, (3
6) is a minimum value selection circuit, which includes arithmetic circuits (35a) to
(35s) is supplied with the outputs of the subtracters (33a) to (33s), respectively, and each arithmetic circuit (35a) to (35s)
The output of is supplied to the minimum value selection circuit (36), and the leading part Ss of the pattern data having the shortest distance to the waveform value of each leading part code pattern is supplied to the symbol synthesis circuit (37). It is supplied to a detection circuit (38). On the other hand, the subsequent part St of the pattern data is supplied from the symbol division circuit (32) to the Viterbi detection circuit (38), the detection output of this detection circuit (38) is supplied to the symbol synthesis circuit (37), and the synthesis circuit ( 38) is led out to the output terminal (300).

Note that the code pattern and pattern distance will be explained in detail in the next section.

G2 Operation of the Embodiment Next, the operation of the embodiment of the present invention will be described with reference to FIGS. 4 to 9.

In this example, the relationship between the recording maximum repetition frequency fo and the pulse width wb at the base level of the isolated pulse is set as shown in the following equation (3), so that the pulse width wb is equal to three times the sampling period. , as shown in Figure 4,
Two sampling points are included within the pulse width wb of each isolated pulse.

Wb = 1/3 ・fo ・−−
-(3) In this specification, the above-mentioned state is referred to as a state in which two intersymbol interference is allowed, and the state in which n sampling points are included within the pulse width wb of each isolated pulse is referred to as a state in which n sampling points are included within the pulse width wb of each isolated pulse. This will be referred to as a state where n intersymbol interferences are allowed.

In this embodiment, the data conversion circuit (1) of the recording system (10) is
3), as shown in Tables 1 and 2 below, 4
4-6 conversion is performed to convert 16 pieces of original data in units of bits into converted data (modulation code) in units of 6 bits.

Table 2 The conversion data in Table 1 is separated into a 3-bit leading part code and a trailing part code, and the leading part code includes 4 data out of the normal 3-bit 8 data patterns. pattern, and these four data patterns are used in most of the subsequent code. As indicated by the * mark in the same table, the subsequent code uses a data pattern that does not exist in the preceding code, and is adjusted so that the DC component of the entire code is [O]. There are some that cannot be fully adjusted and some DC component remains.

In this example, as shown in Table 2, the four modulation codes that have DC components remaining in Table 1 are replaced with modulation codes that have DC components of opposite polarity. A modulation code group is formed.

Then, the DC component adjustment circuit (1) of the data conversion circuit (13)
3C) (see Figure 2), the sum of the frequencies of 0000 and 1000 in the original data of Tables 1 and 2, and 01
11 and 1111, and based on this, adjust the DC component to be [0] by appropriately switching and using both the modulation code groups in Tables 1 and 2. can do.

Table 3 In addition, as shown in Table 3, the above-mentioned DC component is, for example,
This is obtained by replacing each "0" with "-1" in the four data patterns of 001, 011, and 100°110 of the preceding part code.

Also, the intermediate series in the same table are [3/4], [0], [
-3/4] are combined differently, and are formed as follows.

First, let n be the number of intersymbol interferences allowed according to the characteristics of the transmission path, and prepare coefficients for weighting that is distributed linearly decreasing from the center, as shown in Table 4 and Figure 5 below. be done.

Next, this weighting is performed using coefficients for each successive n code among the Ns codes of each replacement data in Table 3.

For example, the code of the first replacement data in Table 3 is ail,
a i2. a i3, and the weighting in Table 4 when two intersymbol interferences are allowed is the coefficient W21.
As W22 (w21 = w22), a il,
The two codes of a i2 and a i2. The two codes of a i3 are weighted, respectively, and the following two sets of values are obtained.

[w21-ail; w22 ・ai2] [w21-
ai2 ; w22 ・ai3] These two weighted codes are added as shown in the following equation (5), and the 1.2nd codes Uil, Ui2 of the 1st intermediate sequence are
is formed.

Similarly, the first and second codes Uj of the third intermediate series
l, Uj2 are formed as shown in equation (6) below.

As is clear from Table 3, the intermediate sequences of the preceding part codes are different from each other, and the code distance V1J, which represents the degree of dissimilarity, is the difference between the th sign of each of the pair of intermediate sequences Ui and Uj. It is defined as the following equation (7) as the sum of absolute values.

Vij=Uil−Ujll+1Ui2−Uj2Σ
Uik-Ujk (7) In this embodiment, a code whose coat distance v1J is twice or more of the weighting reference value is selected as the leading part of the modulation code. This reference value corresponds to the peak value of the impulse response according to the overall transmission characteristics of the recording system (10) and the reproduction system (20).

Note that the predetermined code distance cannot be achieved in the subsequent part code of a data pattern that does not exist in the preceding part code, as shown with an asterisk in Table 1, but as described later, the condition that the preceding code is determined is Under , the modulation code is detected with the same quality as the preceding code, and the predetermined code distance is maintained for the modulation code as a whole.

On the other hand, in the reproduction system (20), in order to cope with the two symbol interferences, the equalization characteristics of the waveform equalization circuit (22) are determined by the following equation (8), which corresponds to class 1 of partial response. The one shown in FIG. 6 is selected.

Pr(1,1)-cos(πf/fo)...-1
8) As a result, the number of levels of the reproduced signal at the output of the waveform equalization circuit (22) is 3, as is well known.
becomes. Then, as shown in Figure 4, Table 1 and 2
A reproduced waveform (code pattern) unique to the data pattern of the preceding part code of each converted data in the table is obtained. This code pattern is r2A at two sampling points within the detection period Tfj where three isolated pulses coexist, when the level of an isolated pulse at a sampling point is A]
, [0].

Each of the three values of [-2A] can be taken.

In addition, the pattern distance representing the degree of dissimilarity of the code patterns P, Q... is a pair of code patterns P, Q...
The playback output levels at the sampling points corresponding to the respective codes are S pk and S qk, respectively.
It is defined as in the following equation (9).

I) pq-Σ 15pk-8qk1...
(9) By adopting the waveform equalization characteristics as shown in Figure 6 and Equation 8), in this embodiment, the codes corresponding to the leading part codes of the converted data in Tables 1 and 2, respectively. The spacing distance Dpq between the patterns P, Q, etc. is more than twice the peak value of the impulse response of the transmission line consisting of the recording system (10) and the reproducing system (20).

Furthermore, since the impulse response waveform is upwardly convex on both sides of the center line, as shown in Figure 4, etc., the pattern distance Dpq in equation (9) is linear as shown in Figure 5. The weighting of the distribution is calculated based on the coefficients, as described in (7) above.
It becomes larger than the code distance V1j in the equation.

In the data detection circuit (30A>) of FIG. 3, the code pattern corresponding to the preceding part code of each converted data as described above is selected as a reference, and each reference value Ro M
It is stored in (34a) to (34s). Subtractor (33
In a) to (33s), these 5 (-4> code patterns) are compared with the preceding pattern in units of 3 bits from the symbol dividing circuit (32).

Based on this comparison output, the arithmetic circuits (35a) to (35
In step s), the pattern distance between the human cover turn and each code pattern is calculated according to the above-mentioned equation (9).

In the minimum value selection circuit (36), each calculation circuit (35
During the output of a) to (35s), one pattern data located on the shortest path JliDmin and any of the eight code patterns is selected, and the maximum likelihood (!, Iaximam Li
kelyhood) detection data, ie, the most likely detection data, are supplied to the symbol synthesis circuit 37) and the Viterbi detection circuit (38).

In the Viterbi detection circuit (38), the selection circuit (36'
Based on the maximum likelihood detection data of the preceding part code Ss supplied from By sequentially selecting the "survival paths", the subsequent part code St, which can be called a convolutional code, is detected completely equivalent to maximum likelihood detection in a state where interference between two codes is allowed.

In this case, the Viterbi detection circuit (38) only needs to process the succeeding part code St of six types of patterns in 3-bit units following the already determined leading part code Ss.
It requires less data processing and can detect errors at a lower error rate.

a successor code St from the Viterbi detection circuit (38);
The leading part code Ss from the minimum value selection circuit (36) is synthesized in the symbol synthesis circuit (37) and sent to the output terminal (3
0o) is derived.

As mentioned above, vector encoding is used for data transformation and detection in this embodiment.

The ternary reproduced output waveform as described above can also be expressed as shown in FIG. 7, and the correspondence with the intermediate series in Table 3 above can be intuitively understood.

For example, in the intermediate sequence of the first replacement data "-1-1+IJ," the first bit [-1/3] corresponds to the reproduction output level [-2A] of the preceding sampling point, and the following bit [0] Playback output level of subsequent sampling points [
0].

In this case, the reproduction system (20) can be synchronized based on zero-crossing points of the reproduced waveform as shown by black circles in FIG.

As shown in the figure, in this embodiment, the codes of the preceding detection area are 4
A type of code has been selected, and a subsequent code sandwiched between such preceding codes cannot achieve a pattern distance of 2A by itself, but under the condition that the preceding code has been determined, it will have the same pattern distance as the preceding code. Detected by quality.

In addition, in this example, as mentioned above, maximum likelihood detection is performed, so the distribution decreases linearly around the three output levels [2A], [0] [-2A] in FIG. The error rate due to characteristic noise is the product of overlapping noise distributions, and corresponds to the area of the portion shown by dense hatching in FIG.

On the other hand, in the case of three-value level detection, the error rate due to distributed noise centered on the three output levels in Figure 7 is shown by the coarse and fine hatches in Figure 8, and the noise distributions overlap each other. Corresponds to the area of the part.

Therefore, the data detection method of this embodiment has improved noise resistance compared to simple three-value level detection.

In this embodiment, for example, 1
At a recording density of 30 kbpi, the symbol error rate is approximately 5 Xl0-', and as shown by the chain line in the same figure, the current DAT (8-10 conversion) has the same error rate at a recording density of just over 7 Q kbpi. The recording density has been significantly improved compared to .

In the above embodiment, during data conversion, 6-focus units of converted data consisting of 3-bit units of leading and trailing parts are determined from the 4-bit unit of original data with an efficiency of [4/6], and at the time of playback, The patterns of the leading and trailing parts of the reproduced signal whose waveforms have been equalized with characteristics that allow interference between two codes, and the code pattern group corresponding to the leading and trailing parts of each converted data are directly or isometrically calculated. Since the data is detected by comparing and selecting an approximate pattern, it is possible to reduce the number of times of data comparison processing and the error rate, and it is also possible to significantly improve the recording density.

G3 Configuration of Other Embodiments Next, other embodiments of the data conversion and detection method according to the present invention will be described with reference to FIGS. 10 to 13.

In a magnetic system to which another embodiment of the present invention is applied, the configuration of the recording system (10) is the same as that shown in FIGS. It will be similar to Figure 1.

FIG. 10 shows the configuration of the main parts of another embodiment of the present invention.

In FIG. 10, parts corresponding to those in FIG. 3 are given the same reference numerals, and some explanations will be omitted.

In FIG. 10, from the symbol dividing circuit (32) of the data detection circuit (30B), each pattern data corresponding to each converted data (symbol data) as shown in Tables 6 and 7 below, St is commonly supplied to subtracters (43a) to (43t) of the second group q.

(44a) to (44t) are the reference values RO of the second group
M, and the waveform values of each subsequent code pattern selected as a reference are stored. Each ROM (44a
) to (44t) are commonly supplied with the output of the synchronization detection circuit (31), and are also supplied with the first minimum value selection circuit (36).
The output of ROM (44a) to (44t) is supplied.
The outputs of the subtracters (43a) to (43t
). The output of each subtractor (43a) to (43t> is the second group of pattern distance calculation circuits (45a) to
(45t) and each arithmetic circuit (45a).
The output of (45t) is supplied to the second minimum value selection circuit (46). The output of this selection circuit (46) is transmitted together with the output of the first selection circuit (36) to the symbol synthesis circuit (37).
). The rest of the configuration is the same as that shown in FIG. 3 above.

The relationship with the pulse width Wb of the isolated pulse is set as shown in the following equation (10), and as shown in FIG. 11, this pulse width wb is equal to four times the sampling period, and the pulse width of each isolated pulse Three sampling points are included within the width wb, allowing three intersymbol interferences.

Wb=1/4 ・fO (10) In addition, in this embodiment, in the data conversion circuit (13) of the recording system (10), each 4-bit unit as shown in Table 5 below is used. Three groups of codes with different DC components are selected as the leading part and the succeeding part, and as shown in Tables 6 and 7 below, 64 original data in 6-bit units are converted into converted data in 8-bit units ( Modulation G4 Operation of Other Embodiments The operation of other embodiments of the invention is as follows.

In this embodiment, in the recording maximum repetition frequency fO and the Table 6 adjustment circuit (13C) (see FIG. 2), the first . By appropriately switching and using the modulation code of the picture table based on the frequency of the second group, the DC component can be adjusted to [01].

Table 7 In the conversion data in Tables 6 and 7, the DC component of the entire code is adjusted by combining the leading part code and the succeeding part code, which have different DC components, but the DC component remains. There are things that are.

Also in this embodiment, the 8th DC component of the data conversion circuit (13)
Table Note that the DC component mentioned above was calculated by replacing each "0" with "-1" in the 12 data patterns of the three groups of codes in Table 5, as shown in Table 8. be.

Also, the intermediate series in Table 8 are [2], [1], [0][
-1] and [-2] are combined differently, and are formed in the following manner as described above.

First, the weighting as shown in Table 4 and FIG. 5 is the coefficient w3L w32. Using w33 (w31=w33), each of the four codes bil to bi4. Three consecutive codes among bjl to bj4 are weighted.

Next, each of the three weighted codes is added as shown in the following equation (11), and l in Table 8 and 1 in the third intermediate series are added.
．． Second code Uil, Ui2; Ujl.

Uj2 are formed respectively.

Uj2=w31bj2+w32bj3+w32bj4
j Then, the code distance VIJ between the first and third intermediate sequences in Table 8 is calculated according to equation (7) above.

In this embodiment as well, codes whose code distance VIJ is twice or more the weighting reference value are selected as all of the leading part and most of the trailing part of the modulation code.

Note that, as shown in Table 5, the following codes of Group C cannot necessarily have a predetermined code distance from the subsequent codes of Groups A and B, but as mentioned above, the preceding code is fixed. Under this condition, the modulated code will be detected with the same quality as the preceding code, and the predetermined code distance will be maintained for the modulated code as a whole.

In the reproduction system (20) of this embodiment, the equalization characteristics of the waveform equalization circuit (22) are as follows to cope with three intersymbol interferences.
The following (1) corresponds to class 2 of partial response.
2) The one shown in the equation and FIG. 12 is selected.

Pr(1,2,1)-cos2(πf/fo) ・
-(12) As a result, in the output of the waveform equalization circuit (22), the number of levels of the reproduced signal becomes 5, as is well known. Also in this example, the sixth,
Code patterns corresponding to the leading and trailing parts of each converted data in Table 7 are obtained.

As illustrated in FIG. 11, in this code pattern, when the peak value of an isolated pulse is B, at two sampling points within a detection period T[j in which four isolated pulses coexist, [2B], [ B], [O], [-Bl, [-2B]
Each of these can take on five different values.

Further, the pattern distance Dpq between the code patterns of groups A and B in the preceding and succeeding parts is more than twice the peak value of the impulse response of the transmission path consisting of the recording system (10) and the reproduction system (20).

As mentioned above, this pattern distance is larger than the code distance v1j of the equation (7) above, and in this embodiment, 8 pieces of converted data in the preceding part and 12 pieces of converted data in the succeeding part as described above. Maximum likelihood detection of input pattern data in units of 4 bits is performed using the code pattern as a reference.

In addition, in this example as well as in the above-mentioned example,
By representing the five-value reproduced output waveform as shown in FIG. 13, the correspondence with the intermediate series in Table 8 can be intuitively understood.

In the above embodiment, during data conversion, 8-bit units of converted data consisting of 4-bit units of leading and trailing parts are determined from the 6-bit unit of original data with an efficiency of [6/8], and at the time of playback, The patterns of the leading and trailing parts of the reproduced signal whose waveforms have been equalized with characteristics that allow interference between three adjacent symbols are directly compared with the code pattern groups corresponding to the leading and trailing parts of each converted data. Since the data is detected by selecting an approximate pattern, it is possible to reduce the number of data comparison processes and the error rate, and it is also possible to significantly improve the recording density.

Effects of the H0 Invention As detailed above, according to the first invention, in a data conversion method for converting original data in units of M bits to be recorded on a medium into converted data in units of N (>M) bits. ,
The N codes of the converted data are divided into Ns codes in the leading part and N codes in the succeeding part.
t symbols, and among the Ns symbols in the leading part or the Nt symbols in the succeeding part, n consecutive symbols corresponding to the allowable number of intersymbol interferences in the transmission path are weighted with a distribution that linearly decreases from the center, When an intermediate sequence is formed by sequentially adding n consecutive weighted codes, the sum of the absolute values of the differences between the codes in the intermediate sequence is the standard value of weighting, as the code of at least the preceding part of the converted data. Since the code pattern with a predetermined number of times or more is selected, the pattern distance between code patterns can be increased, and the error rate during playback can be reduced while using current media and recording/playback devices. Thus, a data conversion method can be obtained which can significantly improve recording density.

Further, according to the second aspect of the invention, in a data detection method for detecting converted data from a reproduced signal of a medium on which converted data in units of N bits are recorded, the medium is adjusted to allow n intersymbol interference. In addition to setting the transmission characteristics of the reproduction system, the data to be detected is set in units of Ns bits in the leading part and in units of Nt bits in the succeeding part, which are obtained by dividing the converted data in units of N bits, and when detecting at least the leading part of the converted data, The pattern of the reproduced signal corresponding to the data to be detected and the code pattern group corresponding to each converted data are compared to select an approximate pattern. , a data detection method that can reliably detect converted data recorded at extremely high density with a low error rate is obtained.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the overall configuration of an embodiment of the data conversion and detection method according to the present invention, FIG. 2 is a block diagram showing the configuration of the main part of an embodiment of the present invention, and FIG. A block diagram showing the configuration of other main parts of an embodiment of this invention, FIG. 4 is a waveform diagram for explaining the operation of an embodiment of this invention, and FIG. 5 is an operation of an embodiment of this invention. FIG. 6 is a diagram showing the characteristics of other essential parts of an embodiment of this invention, and FIGS. 7 and 8 are diagrams for explaining the operation of an embodiment of this invention. FIG. 9 is a schematic diagram illustrating the effects of one embodiment of the present invention, and FIG. 10 is a block diagram showing the configuration of the main parts of another embodiment of the present invention.
1 is a waveform diagram for explaining the operation of another embodiment of the present invention, FIG. 12 is a diagram showing the characteristics of the main part of another embodiment of the present invention, and FIG. 13 is a waveform diagram for explaining the operation of another embodiment of the present invention. A route diagram for explaining the operation of the embodiment, FIG. 14 is a diagram for explaining conventional digital magnetic recording, FIG. 15 is a diagram showing the characteristics of the conventional example, FIGS. 16, 17, and FIG. 18 is a waveform diagram for explaining the present invention, and FIG. 19 is a diagram showing the characteristics of another conventional example. (10) is a recording system, (13) is a data conversion circuit, (13)
a). (13b) is a ROM table, (13e) is a DC component adjustment circuit, (20) is a reproduction system, (22) is a waveform equalization circuit, and (30). (30A), (30B) are data detection circuits, (3
2) is a symbol dividing circuit, (34a) to (34s)
, (44a) to (44t) are reference values R○M, (3
6) and (46) are minimum value selection circuits, and (37) is a symbol synthesis circuit. To the face ○ Ku○ Ro Takukanlt ku ○ ・ ／ ／ ／ ／ ／ ／ ・ ○ 46〇- e) To ○ Sheep Kuichi exchange-1 do e ku w' (Sui fun-n→

Claims (1)

[Claims] 1. The original data in units of M bits to be recorded on the medium is N(>
M) In a method of converting data into converted data in units of bits, the N codes of the converted data are divided into Ns leading parts and Nt trailing parts, and the N codes of the leading part or the trailing part are divided into out of Nt symbols, n equal to the number of intersymbol interferences allowed according to the characteristics of the transmission path.
The weights are weighted to decrease linearly from the standard value at the center of the distribution, and the weighted n consecutive codes are sequentially added to each other in units of Ns-n+1 bits or Nt-n+1. When an intermediate sequence in bit units is formed, the sum of the absolute values of the differences between codes between the intermediate sequences is at least a predetermined times the criterion value for the weighting, as the code of at least the preceding part of the converted data. A data conversion method characterized by selecting items. 2. In a data detection method for detecting the converted data from a reproduced signal of a medium on which N bits of converted data are recorded, the transmission characteristics of the reproduction system of the medium are set to allow n intersymbol interferences. At the same time, the data to be detected is divided into Ns bit units of the leading part and Nt bits of the succeeding part, which are obtained by dividing the converted data in units of N bits, and when detecting at least the leading part of the converted data, the data to be detected is The signal pattern of the reproduced signal corresponding to the data is compared with a plurality of predetermined patterns limited by the conversion data, and an approximate pattern that approximates the signal pattern is selected from the plurality of predetermined patterns to detect data. A data detection method characterized by: