Claims
The invention claimed is:
1. An optical disc device for reading out information by using a PRML procedure, the information being recorded by using codes whose minimum run length is 2T,
the optical disc device comprising:
means of binarizing each of readout signal waveforms acquired from the optical disc by using the PRML procedure so as to obtain each of first binarized bit arrays;
means of generating a second binarized bit array and a third binarized bit array by shifting an edge of interest in each of the first binarized bit arrays by 1T leftward and rightward, respectively, wherein the second binarized bit array and the third binarized bit array are generated by calculating the first binarized bit array using exclusive OR (XOR) mask bit arrays, and the mask bit arrays include two kinds or more, at least one of which has a hamming distance of 2;
means of generating a first target signal waveform, a second target signal waveform and a third target signal waveform corresponding to each of the first binarized bit arrays, each of the second binarized bit arrays and each of the third binarized bit arrays, respectively;
means of calculating a first value equivalent to a difference between a Euclidean distance from the second target signal waveform to the readout signal waveform and a Euclidean distance from the first target signal waveform to the readout signal waveform, and a second value equivalent to a difference between a Euclidean distance from the third target signal waveform to the readout signal waveform and a Euclidean distance from the first target signal waveform to the readout signal waveform;
means of calculating a shift evaluation value of the edge of interest by using a differential value between the first value and the second value; and
means of using a pulse string including (N−1) pulses for forming a recorded mark having a length of NT, where N is an integer, and of adjusting a starting position of a last pulse in the pulse string so as to minimize the shift evaluation value of the edge.
2. The optical disc device according to claim 1 , wherein
the adaptive equalization procedure sets to each of tap coefficients C n an average value of a pair of tap coefficients a n and a N1n located symmetrically with each other relative to the time axis, the tap coefficients a n are updated by using a LMS method.
3. A method for adjusting a recording condition on an optical disc for writing information by using codes whose minimum run length is 2T, and for reading out the information by using an adaptive equalization procedure and a PRML procedure,
the PRML procedure including a PR(1, 2, 2, 2, 1)ML method,
the method for adjusting a recording condition comprising:
a step of binarizing each of readout signal waveforms acquired from the optical disc by using the PRML procedure so as to obtain each of first binarized bit arrays;
a step of generating a second binarized bit array and a third binarized bit array by shifting an edge of interest in each of the first binarized bit arrays by 1T leftward and rightward, respectively;
a step of generating a first target signal waveform, a second target signal waveform and a third target signal waveform corresponding to each of the first binarized bit arrays, each of the second binarized bit arrays and each of the third binarized bit arrays, respectively;
a step of expressing
xL by
xL
=
1
2
(
1

ED
(
L
,
W
)

ED
(
T
,
W
)
ED
(
T
,
L
)
)
,
(
D1
)
xR by
xR
=
1
2
(
1

ED
(
R
,
W
)

ED
(
T
,
W
)
ED
(
T
,
R
)
)
,
(
D2
)
and an extended edge shift D by
D
=
xR

xL
2
where
W represents the readout signal waveform, T represents the first target signal waveform, L represents the second target signal waveform, R represents the third target signal waveform, ED(a, b) (a and b are any one of W, T, R, L) represents a Euclidean distance among W, T, R and L, and xL represents an evaluation value for an error caused when the edge of interest is shifted leftward, xR represents an evaluation value for an error caused when the edge of interest is shifted rightward, and the extended edge shift D represents amount of an edge shift of the edge of interest, and
at least one of a Hamming distance between the second target signal waveform L and the first target signal waveform T, and a Hamming distance between the third target signal waveform R and the first target signal waveform T is 2 or greater, and the second target signal waveform L, the first target signal waveform T, and the third target signal waveform R satisfy run length restriction, and
calculating a shift evaluation value that is an statistical average A of the extended edge shifts Ds; and
a step of using a pulse string including (N−1) pulses for forming a recorded mark having a length of NT, where N is an integer, and of adjusting a starting position of a last pulse in the pulse string so as to minimize the shift evaluation value of the edge.
4. The method for adjusting a recording condition according to claim 3 , further comprising:
a step of classifying and calculating, on a basis of the readout signal waveform, the edge shift evaluation value corresponding to an end edge of the recorded mark into 3×4 combination patterns constituted by three recorded mark lengths of 3T, 4T, and 5T or greater and four succeeding space lengths of 2T, 3T, 4T and 5T or greater; and
a step of determining whether each of the shift evaluation value of the edges classified and calculated is minimum or not,
wherein
a last pulse starting position is adjusted until each of the shift evaluation value of the edges classified and calculated becomes minimum.
5. The method for adjusting a recording condition according to claim 3 , wherein
an adjustment unit of adjusting the last pulse starting position is set at 1/16 of a channel bit period.
6. An optical disc device for reading out information by using a PR(1, 2, 2, 2, 1)ML procedure, the information being recorded by using codes whose minimum run length is 2T,
the optical disc device comprising:
means of binarizing each of readout signal waveforms acquired from the optical disc by using the PRML procedure so as to obtain each of first binarized bit arrays;
means of generating a second binarized bit array and a third binarized bit array by shifting an edge of interest in each of the first binarized bit arrays by 1T leftward and rightward, respectively;
means of generating a first target signal waveform, a second target signal waveform and a third target signal waveform corresponding to each of the first binarized bit arrays, each of the second binarized bit arrays and each of the third binarized bit arrays, respectively;
means of expressing
xL by
xL
=
1
2
(
1

ED
(
L
,
W
)

ED
(
T
,
W
)
ED
(
T
,
L
)
)
,
(
D1
)
xR by
xR
=
1
2
(
1

ED
(
R
,
W
)

ED
(
T
,
W
)
ED
(
T
,
R
)
)
,
(
D2
)
and an extended edge shift D by
D
=
xR

xL
2
where
W represents the readout signal waveform, T represents the first target signal waveform, L represents the second target signal waveform, R represents the third target signal waveform, ED(a, b) (a and b are any one of W, T, R, L) represents a Euclidean distance among W, T, R and L, and xL represents an evaluation value for an error caused when the edge of interest is shifted leftward, xR represents an evaluation value for an error caused when the edge of interest is shifted rightward, and the extended edge shift D represents amount of an edge shift of the edge of interest, and
at least one of a Hamming distance between the second target signal waveform L and the first target signal waveform T, and a Hamming distance between the third target signal waveform R and the first target signal waveform T is 2 or greater, and the second target signal waveform L, the first target signal waveform T, and the third target signal waveform R satisfy run length restriction, and
calculating a shift evaluation value that is an statistical average Δ of the extended edge shifts Ds; and
means of using a pulse string including (N−1) pulses for forming a recorded mark having a length of NT, where N is an integer, and of adjusting a starting position of a last pulse in the pulse string so as to minimize the shift evaluation value of the edge.
7. A method for recording data on an optical disc with codes whose minimum run length is 2T using a PRML procedure, including a step for adjusting a recording condition,
the method for adjusting a recording condition comprising:
a step of binarizing each of readout signal waveforms acquired from the optical disc by using the PRML procedure so as to obtain each of first binarized bit arrays;
a step of generating a second binarized bit array and a third binarized bit array by shifting an edge of interest in each of the first binarized bit arrays by 1T leftward and rightward, respectively;
binarized bit arrays, respectively,
a step of expressing
xL by
xL
=
1
2
(
1

ED
(
L
,
W
)

ED
(
T
,
W
)
ED
(
T
,
L
)
)
,
(
D1
)
xR by
xR
=
1
2
(
1

ED
(
R
,
W
)

ED
(
T
,
W
)
ED
(
T
,
R
)
)
,
(
D2
)
and an extended edge shift D by
D
=
xR

xL
2
where
W represents the readout signal waveform, T represents the first target signal waveform, L represents the second target signal waveform, R represents the third target signal waveform, ED(a, b)(a and b are any one of W, T, R, L) represents a Euclidean distance among W, T, R and L, and xL represents an evaluation value for an error caused when the edge of interest is shifted, xR represents an evaluation value for an error caused when the edge of interest is shifted rightward, and the extended edge shift D represents amount of an edge shift of the edge of interest, and
at least one of a Hamming distance between the second target signal waveform L and the first target signal waveform T, and a Hamming distance between the third target signal waveform R and the first target signal waveform T is 2 or greater, and the second target signal waveform L, the first target signal waveform T, and the third target signal waveform R satisfy run length restriction, and
calculating a shift evaluation value that is an statistical average of the extended edge shifts Ds;
a step of using a pulse string including (N−1) pulses for forming a recorded mark having a length of NT, where N is an integer, and of adjusting a starting position of a last pulse in the pulse string so as to minimize the shift evaluation value of the edge; and
a step of recording data using the pulse string with said adjusting starting position of the last pulse.
CROSS REFERENCE TO RELATED APPLICATION
U.S. patent application Ser. No. 12/569,021 is a copending application of this application, the content of which is incorporated herein by crossreference.
FIELD OF THE INVENTION
This invention relates to a method for adjusting recording condition in recording information in an optical disc medium on which recorded marks are formed that have a physical property different from the physical property of that part of the optical disc medium which is other than the recorded marks, an optical disc device using the adjusting method, and a method for recording information.
BACKGROUND OF THE INVENTION
Optical disc media now widely available include CDR/RW, DVDRAM, DVD±R/Rw, BD, etc., and some of them have two data layers. Optical disc devices adapted for record and playback with those optical disc media mentioned above, i.e. socalled DVD super multidrives, are now widely in use. In the near future, highperformance disc drives adapted to blue ray discs (hereafter referred to simply as BDs) will come to be widely used. Further, there's need for optical discs having still higher capacity.
The increase in the drive speed of optical disc drive and in the density of information recorded in the optical disc has necessitated the introduction of a technique in which the binarization of reproduced signal is performed by Partial Response Maximum Likelihood (PRML) reproduction procedure. As one of the PRML procedures is known the adaptive PRML procedure (or system) or compensatory PRML procedure (or system) which can adaptively change the target signal level in accordance with the reproduced signal. The nonpatent document 1, “Journal C of Institute of Electronics, Information and Communication Engineers, Vol. J90C, p. 519 (2007)” discloses the fact that a device adapted to BDs can achieve highdensity recording equivalent to 35 GB by using such a PRML procedure wherein the asymmetry of reproduced signal and the thermal interference at the time of recording can be compensated. It is pointed out in the document that reproduction performance under the condition for highdensity recording is higher for larger constraint length (bit length representing ‘class’). In order to obtain the best result from binarization, an automatic equalizer that makes minimum the RMS error between reproduced signal and the target signal for the PRML index, is installed in an optical disc device provided with such a PRML procedure. In general, such an automatic equalizer is installed as a Finite Impulse Response (FIR) filter having its tap factors variable.
The increase in the recording density in an optical disc leads to the decrease in the size of each recorded mark as compared with the size of the light spot, and therefore results in the reduction of the amplitude of the obtained reproduced signal. The resolving power of the light spot is determined by the wave length λ and the aperture number NA of the objective lens. Accordingly, if the length of the recorded mark having the shortest run length becomes equal to or less than λ/4NA, the amplitude of the signal corresponding to the repeated series of such shortest recorded marks is reduced to zero. This phenomenon is known as “optical cutoff”, and may occur when λ/4NA≅119 nm in the case of BDs. In the case of BDs, an attempt to attain a recording capacity of more than about 31 GB with the track pitch kept constant, causes the amplitude of the signal corresponding to the repeated series of recorded marks having the shortest run length of 2T to be reduced to zero. It is therefore indispensable to use a PRML procedure in order to acquire a satisfactory reproduction performance under such a highdensity condition.
When desired information is recorded in a recordable optical disc, the state of crystallization, for example, in the recording film is changed by using pulsed laser light whose intensity is modulated (hereafter referred to as ‘recording pulse’). Materials for such recording films include phasevariable substances, organic pigments, certain kinds of alloys and oxides, all of which are well known and widely used. In the mark edge coding method for use with CDs, DVDs and BDs, code information is determined depending on the positions of leading and trailing edges. Regarding the recording pulses, the positions and widths of the first pulse that mainly determines the condition for forming the leading edge of a recorded mark and the last pulse that mainly determines the condition for forming the trailing edge of the recorded mark, are important to maintain the quality of recorded information in a good condition. Therefore, it is customary with recordable optical discs to use “adaptive recording pulses” which can adaptively change the positions or widths of the first and last pulses in accordance with the length of each recorded mark and the lengths of the spaces that immediately precedes and follows the recorded mark.
FIG. 43 illustrates recording pulse waveforms in an “N−1 write strategy” that is one of recording methods used for BDs. In this write strategy, a mark having a length of NT is written by using (N−1) pulses. FIG. 43 illustrates pulse strings for recorded marks having lengths of 2T to 5T among marks having lengths of 2T to 9T (T denotes a “channel bit length”) used for codes of BDs. The first maximum power pulse in a pulse string is referred to as a “first pulse” and the last maximum power pulse in a pulse string is referred to as a “last pulse”. Multiple maximum power pulses between the first pulse and the last pulse are referred to as “multi pulses”, a 4T mark includes one multi pulse, and the number of the multi pulses increase by one as the mark length increases by 1T. A pulse string for a mark having a length of 2T includes only the first pulse as the maximum power pulse, a pulse string of a 3T mark includes the first pulse and the last pulse as the maximum power pulses, and both the marks include no multi pulse. A pulse immediately following the last maximum power pulse in a pulse string is referred to as a “cooling pulse”.
There are four output power levels in a laser beam: a write power P W , a space power (erase power) P S , a bias power P BW , and a cooling power P C . The write power P W is the maximum power level in a pulse string, and is a power level used for a first pulse, each multi pulse and a last pulse. This power level is used for causing a state change to a recording film by applying energy to the recording film. The space power P S is a power level used for irradiating a portion (space) to be located between marks, and is used chiefly for preheating to form a succeeding mark in an application of a writeonce disc, and is used chiefly for erasing a mark, in which a space allowing a direct overwrite is created in an application of a rewritable disc using a phase change recording film. The cooling power P C is a power level for a cooling pulse, and is used chiefly for blocking thermal diffusion and reducing thermal interference to a succeeding mark recorded portion in an application of a writeonce disc, and is used chiefly for forming an amorphous mark by rapidly cooling a recording film after heating in an application of a rewritable disc. Each type of the power levels has the same value regardless of the mark length.
Parameters regarding pulse timing include a start edge position dTtop of a first pulse, a duration Ttop of the first pulse, a duration TMP of each multi pulse, a duration TLP of a last pulse and an end edge position dTS (dTE) of a cooling pulse. Both dTtop and dTS (dTE) are defined on the basis of an NRZI channel bit signal of written data, as shown in FIG. 43 . Each adjustment unit of the parameters is set at 1/16 of the channel bit period.
Among the pulse parameters, dTtop and Ttop that chiefly define a formation condition for a front edge of a recorded mark, and TLP and dTS (dTE) that chiefly define a formation condition for an end edge of the recorded mark are important for preferably maintaining quality of written information. In BDs, thus, an adaptive write pulse is used for adaptively changing the above described parameters depending on the length of a recorded mark and the length of preceding or succeeding space thereof. Each value of dTtop and Ttop is classified and specified into combination patterns based on the length of a recorded mark and the length of space immediately before the recorded mark (preceding space), and each value of TLP and dTS (dTE) is classified and specified into combination patterns based on the length of a recorded mark and the length of space immediately after the recorded mark (succeeding space). Each value of TMP not illustrated is not classified based on the length of a mark or the length of space thereof, and the same value is specified to every mark having a length of 4T or greater.
Under such a highdensity recording condition as described above, since the size of each formed recorded mark becomes very small, it is necessary to choose the condition for radiating the recording pulses (hereafter referred to as “recording condition”) with a higher precision than conventional. On the other hand, in an optical disc device, the shape of the light spot varies depending on the wavelength at the light source, wave front aberration, focusing condition, the tilt of disc, etc. Further, since the ambient temperature and the aging effect change the impedance and the quantum efficiency of the semiconductor laser device, the shapes of the recoiling pulses change accordingly. The technique for invariably obtaining the best recording condition in response to the shapes of light spots and the shapes of the recording pulses both of which fluctuate depending on environments and devices, is usually called “test writing”. Such a technique for adjusting the recording condition by using the test writing will become more and more important with the requirement for further increasing recording density.
Adjusting techniques for recording condition are classified roughly into two categories: one method uses bit error rate or byte error rate as index and the other utilizes statistical index such as jitter. The former pays attention to an event that occurs with a small probability with respect to recorded data and the latter is concerned with the average quality of recorded data. Regarding writeonce optical discs, for example, in the case where data are recorded in and reproduced from plural locations in the disc with the recording condition varied, even the best recording condition for the former method may cause a large bit error or byte error if fingerprints overlie the recorded data. Therefore, the former method should not be selected in this case. The best recording condition should be such that the average quality of the data recorded under such a recording condition is optimal. It can therefore be said that the method using statistical index is preferable for storage media such as optical discs, which are vulnerable to material flaws, fingerprints or dust.
Methods corresponding to PRML procedure for statistically evaluating the quality of recorded data are disclosed in, for example, “Jpn. J. Appl. Phys. Vol. 43, p. 4850 (2004)” (nonpatent document 2), JP 2003141823 A (patent document 1), JP 2005346897 A (patent document 2), JP 2005196964 A (patent document 3), JP 2004253114 A (patent document 4), and JP 2003151219 A (patent document 5).
The patent document 1 discloses the technique wherein use is made of the certainty Pa corresponding to the most likelihood state shift array and the certainty Pb corresponding to the secondary likelihood state shift array so that the quality of reproduced signal is evaluated on the basis of the distribution of Pa−Pb. The nonpatent document 2 discloses a technique wherein the value obtained by subtracting the Euclidean distance between two target signals from the absolute value of the difference between the Euclidean distance (corresponding to Pa) between the target signal representing the binary bit array (corresponding to the most likelihood state shift array) derived from the reproduced signal and the reproduced signal, and the Euclidean distance (corresponding to Pb) between the target signal representing the binary bit array (corresponding to the secondary likelihood state shift array) derived through a singlebit shift of the interested edge and the reproduced signal, is defined as MLSE (Maximum Likelihood Sequence Error), and the recording condition is adjusted in such a manner that the average value of the distribution of MLSEs is reduced to zero for every recorded pattern.
The patent document 2 discloses a technique wherein edge shift is specifically noted: a virtual pattern having a run length of 1T is used as an error pattern for showing that the edge of reproduced signal shifts to the right or left; the amount of edge shift is obtained by calculating the difference between sequence errors having plus or minus sign depending on the direction in which the edge shift occurred; and the recording condition is so adjusted as to cause the amount of edge shift to approach zero. In this case, the evaluating index is called “VSEAT (Virtual state based Sequence Error for Adaptive Target) index”. The patent documents 3 and 4 disclose a technique wherein the difference between the Euclidean distance between reproduced signal and correct pattern and the Euclidean distance between reproduced signal and error pattern, is calculated by using a table containing the combinations of correct patterns and error patterns corresponding the correct patterns; and the Simulated bit Error Rate (SbER) is obtained from the average and standard deviation of the Euclidean distance differences.
The patent document 5 discloses a technique wherein, on the basis of the difference between the Euclidean distance between reproduced signal and correct pattern and the Euclidean distance between reproduced signal and error pattern, the error probabilities corresponding respectively to the case where the interested edge has shifted to the left and to the case where it has shifted to the right, are obtained; and the recording condition is so adjusted as to make the probabilities corresponding to the two cases equal to each other. Accordingly, use is made of a preselected reproduced signal, a first pattern whose wave pattern corresponds to that of the preselected reproduced signal, and an arbitrary pattern (a second or a third pattern) whose wave pattern corresponds to that of the preselected reproduced signal but which is different from the first pattern. First, the distance difference D=Ee−Eo between the distance Eo between the reproduced signal and the first pattern and the distance Ee between the reproduced signal and the arbitrary pattern, is obtained. Secondly, the distribution of the distance differences Ds with respect to plural samples of reproduced signals is obtained. Thirdly, the quality evaluation parameter (M/σ) is determined on the basis of the ratio of the average M of the obtained distance differences Ds to the standard deviation σ of the obtained distribution of the distance differences Ds. And finally, the quality of reproduced signal is assessed from the evaluation index value (Mgn) represented by the quality evaluation parameters.
PRIOR ART DOCUMENTS
Patent Documents
Patent Document 1: JP 2003141823 A
Patent Document 2: JP 2005346897 A
Patent Document 3: JP 2005196964 A
Patent Document 4: JP 2004253114 A
Patent Document 5: JP 2003151219 A
NonPatent Documents
Nonpatent Document 1: Journal C of Institute of Electronics, Information and Communication Engineers, Vol. J90C, p. 519 (2007)
Nonpatent Document 2: Jpn. J. Appl. Phys. Vol. 43, p. 4850 (2004)
SUMMARY OF THE INVENTION
Problems to be Solved by the Invention
The most likelihood state transition array and the secondary likelihood state transition array described in the patent document 1 are the same as the correct and error patterns described in the patent document 3, in that they are both the target bit arrays for measuring distances to reproduced signals. The patent documents 2 and 5 discloses three target bit arrays, which are all the same in meaning. These target bit arrays will hereafter be called an “evaluation bit array” collectively. Moreover, this invention basically aims to provide BD systems having a high capacity of more than 30 GB, and the following description will be made on the assumption that the shortest run length for modulation code is 2T.
As described in the nonpatent document 1, a PRML system having a constraint length of 5 or greater is preferable to achieve highdensity recording. As described above, when recording density along the track (or linear recording density) is increased, the standard optical requirement for BD (wave length: 405 nm; aperture number of objective lens: 0.85) will cause the amplitude of the signal repeated every 2T period to be reduced to zero. In such a case, it is well known that the PRML system with PR(1, 2, 2, 2, 1) characteristic in which the amplitudes of continuous 2T signals are reduced to zero, is suitable. The patent documents 3 and 4 disclose the method, adapted for PR(1, 2, 2, 2, 1) characteristic, calculating SbER (Simulated bit Error Rate) which is used for evaluating the quality of reproduced signal. This SbER method utilizes the binarized bit arrays (correct patterns) and the secondary likelihood evaluation bit arrays (error patterns) such as a bit array having Hamming distance of 1 from the correct pattern (an edge shift), a bit array having Hamming distance of 2 from the correct pattern (shift of 2T data) and a bit array having Hamming distance of 3 from the correct pattern (shift of 2T2T data); regards the distribution of such patterns as Gaussian distribution; and estimates the bit error rate from the average and the standard deviation by using the error function.
Description will be made below of the performance required for the technique for adjusting the highprecision recording condition needed for realizing an optical disc system having a recording capacity of 30 GB or greater on the basis of the BD standard. Such a technique requires, with respect to at least the quality of the data recorded according to the result of adjustment, (1) that SbER and bit error rate should be sufficiently small and (2) that the data recorded on a disc with a disc drive should give sufficiently small SbER and bit error rate when they are reproduced from the same disc with another disc drive. The required performance (1) given above seems to be a matter of course, but the required performance (2) is specifically sought for since the disc drive must reproduce the data recorded with another disc drive. Therefore, it cannot be said that the adjusting method for recording condition which cannot meet the above described required performances (1) and (2) is suitable for the highdensity optical disc system.
From the viewpoint of the two required performances described above, description will be made below of the problems associated with the conventional techniques and their combinations.
First, explained are various events caused when highdensity recording and reproducing having recording capacity of 30 GB or greater per disc surface is performed by way of experiments and simulations in which linear recording density is increased.
FIG. 2 graphically shows an experimental result that illustrates the relationship between bit error counts and recording power measured by using a three layer wrightonce optical disc which was produced for testing purposes. The thin layers serving as recording medium on the test disc were made of Gebased chemical compound. The gaps between the top and middle layers and between the middle and bottom layers were 14 μm and 18 μm, respectively. The thickness of the transparent covering layer was 100 μm, measured from its top surface to its bottom surface contacting the bottom layer. The track pitch was 320 nm. The recording/reproducing condition was such that data transfer speed is double that for BD and that the width (1T) of the detecting window was set equal to about 56 nm. Accordingly, the desired recording density was designed to be 33 GB. Laser beams of ordinary multipulse type modulated for three power levels, i.e. peak power, assist power and bottom power, were used as the sources of recording pulses. The processing system for reproduced signal comprised an 8bit A/D converter, an automatic equalizer having 21 taps, and a Vitervi decoder having PR(1, 2, 2, 2, 1) characteristic. The minimum value for BER was less than 10 −5 for each of the three layers. The peak power values for layers L0, L1 and L2, which caused BER to become minimal, was 13.5 mW, 15.5 mW and 11.5 mW, respectively. FIG. 2 graphically shows the relationship between recording power and bit error counts while the recording power was changed with the ratio among the three power levels maintained constant. This graph reflects the case where an edge shifts and also 1 to 4 consecutive 2T collectively shift (slip). It is seen from FIG. 2 that the bit error counts with respect to not only the edge shift but also the collective shifts are likewise larger than anticipated, in response to the deviation of the recording power. This result is due to the fact that the amplitude of the 2T2T signal is 0, and that Euclidean distance of “12” for the collective shift of consecutive 2T is smaller than Euclidean distance of “14” for the edge shift in the case of PR(1, 2, 2, 2, 1) characteristic.
FIG. 3 graphically shows, as a result of simulation, the relationship between SNR and SbER. In this simulation, the impulse response obtainable when recorded marks are reproduced was obtained by a linear diffraction simulator, and the reproduced signal obtainable when the recording is ideally performed was calculated by convolving the recorded bit array and the impulse response. Noise was added as white noise, and SNR was determined as the ratio of half the peak value of the 8Trepetitive signal to the standard deviation of the noise. The desired bit error rate and SbER were obtained through processing with a reproduced signal processing system having PR(1, 2, 2, 2, 1) characteristic. The patent document 3 discloses the evaluation pattern corresponding to the case where up to two continuous 2T signals continue. In this simulation, up to six continuous 2T signals were caused to continue. This is an extension (Hamming distances of 1 to 7) of the disclosed case. There are 18 evaluation patterns per Hamming distance so that the total number of the evaluation patterns is 252. It is apparent from FIG. 3 that the values of SbER remain almost constant when the number of continuous 2T signals is equal to or greater than 2 (Hamming distance of 3). Although this result seems to be contradictory to the experimental result shown in FIG. 2 , it is not the case. The definition in the calculation of SbER permits the estimation of bit error rate with the existence probability of evaluation pattern taken into consideration, and even an evaluation covering up to two continuous 2T signals can estimate the overall bit error rate.
FIG. 4 graphically shows an experimental result indicating the relationship between bit error rate and SbER. An experiment was done in such a manner that recording along five tracks is continued so as to include the influence of crosstalk in the L0 layer and that various recording/reproducing stresses are imposed on the track in the center. The stresses include the radial tilt (Rtilt) of disc, the tangential tilt (Ttilt) of disc, the aberration of focus (AF), the spherical aberration (SA) due to improper adjustment of the optical head beam expander, and the change in the recording power (Pw). Regarding the radial tilt, the result with respect to the L2 layer is also shown. It is apparent from FIG. 4 that there is a good correlation between bit error rate and SbER. The reason why there is a large scatter of data points around the bit error rate of 10 −5 is ascribed mainly to the flaws in the recording material used in this experiment.
As is apparent from the result of the experiment and simulation, it is necessary that under the requirement for the highdensity recording and reproduction capable of achieving a recording capacity of 33 GB per disc surface, the evaluation of bit errors covering not only an edge shift (Hamming distance of 1) but also up to at least two continuous 2T signals (Hamming distance of 3) should be performed. The method of evaluating the quality of reproduced signals, which concentrates only on the edge shift, cannot be said to develop a sufficient correlation between bit error rate and SbER.
The distribution of Euclidean distance differences accompanying highdensity recording will now be described. The term “Euclidean distance difference” used in this specification signifies the value resulting from subtracting the Euclidean distance between reproduced signal and correct target signal from the Euclidean distance between reproduced signal and error target single. This Euclidean distance difference is defined as Pa−Pb in the patent document 1 and as D in the patent documents 3 and 4. Here, in order to consider an ideal recording state, such a simulation as described above was used. SNR was set to 24 dB, and the distribution of the Euclidean distance differences covering up to two continuous 2T signals was calculated while recording density was changed within a range of 25 to 36 GB per disc surface (T=74.5 nm˜51.7 nm). The configuration of the reproduced signal processing system is as described above. FIG. 5 shows the result of such calculations. The obtained distributions are sometimes called the “SAM distribution”. As described above, with PR(1, 2, 2, 2, 1) characteristic, the ideal Euclidean distance (=14) for an edge shift is different from the Euclidean distance (=12) for the shift of a 2T signal and the Euclidean distance (=12) for the shift of two continuous 2T signals. And in order to denote them collectively, each Euclidean distance difference was normalized by dividing it with the ideal Euclidean distance. In FIG. 5 , the statistical probability corresponding to the case where the distance difference becomes zero (at the leftmost end) or negative, gives the bit error rate. As seen in FIG. 5 , it is apparent that the distribution spreads wider with the improvement in recording density though the SNR is kept constant. This tendency indicates the increase in error rate with the increase in recording density, and therefore is quite reasonable. The average value of the distribution (nearly equal to the peak value) for an edge shift remains the same at the position of the horizontal axis of near 1 (=ideal Euclidean distance) even when the recording capacity increases. However, the average value of the distribution, i.e. the peak value, for the shift of a single 2T signal or two continuous 2T signals moves toward zere when the number of shifts and the recording capacity increase. The cause of this phenomenon is considered to depend on the processing capacity of the automatic equalizer used. As described above, the automatic equalizer works in such a manner that the RMS error between reproduced signal and correct target signal is minimized. Since the sampling interval is 1T that is a finite value, the calculation of discreet frequency characteristics can only be performed up to half the sampling frequency according to the Sampling Theorem. In this way, since the filter characteristic obtainable with an automatic equalizer is restricted as described above, the amplitudes of the higher harmonic components of the reproduced signal become large for a recorded pattern in the region where a long continuation of 2T marks (spaces) occurs. As a result, it is considered that the upper limit of the processing capacity of the automatic equalizer is approached so that the deviation from the ideal Euclidean distance becomes large. As described later, the phenomenon that the peak value (or average value) of the distribution of Euclidean distance differences shifts with the increase in recording density, toward the direction in which the peak value tends to be smaller than the ideal Euclidean distance, is an important event concerning the technique for adjusting recording condition. It is noted here that none of the above quoted documents have a description of this phenomenon.
On the basis of the above described experiment and simulation, from the viewpoint of the two required performances mentioned above, problems arising from each or a combination of the conventional techniques described above are summed up as follows.
(1) Method Disclosed in the NonPatent Document 2
The nonpatent document 2 discloses the technique wherein edge shifts are adjusted in such a manner that the average value of the distribution of the Euclidean distance differences becomes equal to the ideal Euclidean distance, on the basis of the technique disclosed in the patent document 1. The “Expression (1)” cited in the nonpatent document 2 defines a specific edge shift MD as a quantity given by the following expression (1).
MD
=
∑
1
4
(
X

P
A
)
2

∑
1
4
(
X

P
B
)
2

d
min
(
1
)
In the above expression, X denotes the level of the reproduced signal; P A and P B are the target signal levels corresponding to the binarized (i.e. binary) bit array (most likelihood state transition array) and a bit array with a onebit edgeshift (secondary likelihood state transition array), respectively; and d min is the Euclidean distance corresponding to the edge shift. By way of supplement, according to the result shown in FIG. 5 , this method is one that corresponds to adjusting the recording condition in such a manner that the distribution of edge shifts takes the peak value at the ideal Euclidean distance (=1). On the other hand, FIG. 3 illustrates the fact that under the condition for highdensity recording the correlation with SbER (or bit error rate) is not sufficient if edge shifts alone are taken into consideration. Consequently, it is understood from the viewpoint of the required performance (1) mentioned above that this method, which takes edge shifts alone into consideration, is not satisfactory under the condition for highdensity recording. Further, the “Table 2” in the nonpatent document 2 reveals that there exists no adjusting index at the location where a 2T mark and a 2T space are juxtaposed to each other, that is, the leading edge (Tsfp(2s, 2m)) of a 2T mark following a 2T space and the trailing edge (Telp(2s, 2m)) of a 2T mark followed by a 2T space. From this point of view along with the result shown in FIG. 2 , it cannot be said that this method is satisfactorily applied to a case of highdensity recording condition in which the error for a 2T mark or space are considerable.
(2) Method Disclose in the Patent Document 2
This method disclosed in the patent document 2, too, takes edge shifts alone into consideration to obtain the index for the adjustment of recording. However, if a virtual 1T mark or space is introduced, the adjustment of recording becomes possible also for the location where two continuous 2T patterns continue. Notwithstanding this, the method takes edge shifts alone into consideration and therefore cannot be said to develop a sufficient correlation with SbER (or bit error rate). Accordingly, this method does not prove to be satisfactory, either, from the viewpoint of the required performance (1) mentioned above.
(3) Method Disclose in the Patent Document 5
According to the method disclosed in the patent document 5, even error bit arrays are so selected as to satisfy the restriction on run length so that high correlation between index and SbER (or bit error rate) can be developed for not only the case of an edge shift but also the case where two 2T patterns shift continuously. This method, therefore, can be said to be an excellent method. According to this method, in order to adjust the recording condition involving a 2T mark, the Hamming distance between an error bit array to be evaluated and a correct bit array takes different values according as the interested mark edge shifts leftward or rightward, as shown in FIG. 3 of the patent document 5. For instance, let Tsfp(3s, 2m) be taken as an example in accordance with the notation employed in the nonpatent document 2. Then, those disclose bit arrays are as follows.
TABLE 1
↓ interested edge
leftshifted bit array
1 1 1 0 0 1 1 1 0 0 0 0
correct bit array
1 1 1 0 0 0 1 1 0 0 0 0
rightshifted bit array
1 1 1 0 0 0 0 1 1 0 0 0
In the case where the partial response characteristic is PR(1, 2, 2, 2, 1), the Hamming distance and the Euclidean distance, between the correct bit array and the leftshifted bit array, are 1 and 14, respectively, whereas the Hamming distance and the Euclidean distance, between the correct bit array and the rightshifted bit array, are 2 and 12, respectively. As shown in FIG. 5 , the difference in Hamming distance causes the difference in the values for the average and the standard deviation of distributions. In order to overcome this problem, the patent document 5 introduces the concept of SbER, estimates their error probabilities by using the error function, and specifies the adjustment target as the condition that their error probabilities are equal to each other. According to this method, it is considered possible to determine such a recording condition as to minimize SbER (or bit error rate). On the other hand, the result of simulation as shown in FIG. 5 and described above corresponds to the case where recorded marks are formed in the ideal condition (without any edge shift). As seen in FIG. 5 , the change in Hamming distance causes the change in the central value and the standard deviation. Therefore, according to the method disclosed in the patent document 5, the condition for forming recorded marks must be adjusted in such a manner that the error probabilities for three distributions (probabilities for which Euclidean distance becomes 0) are rendered equal to one another. From the viewpoint of the required performance (2) mentioned above, that is concerned with the warranty of disc compatibility, there is room for question of whether this method is an ideal one for adjusting recording condition for highdensity optical discs. For the purpose of quantitative evaluation on this point, analyses were made by using such a simulation as described above.
The extension of concept was introduced to define the amount of edge shift to be detected by the method according to the patent document 5. According to the expression (13) in the patent document 5, the quantity Ec equivalent to the edge shift is defined with the following expression (2).
E c =(σ 3 *M 2 +σ 2 *M 3 )/(σ 2 +σ 3 ) (2)
In the above expression, M 2 and M 3 , and σ 2 and σ 3 represent the averages and the standard deviations of distributions of the Euclidean distance differences calculated when the interested edge is shifted by one bit to the left and to the right, respectively. The result shown in FIG. 5 was obtained by normalizing the two distributions with the ideal Euclidean distance, as described above. In like manner, if the ideal Euclidean distance is assumed to be equivalent to 1T, the edge shift Ec′ in the direction of time axis can be calculated from the amount Ec equivalent to the edge shift by using M 2 , M 3 , σ 2 and σ 3 normalized with the ideal Euclidean distance.
FIG. 6 shows the distributions obtained through simulation. It is evident from FIG. 6 that the same result as what is schematically shown in FIG. 6 of the patent document 5 has been obtained. FIG. 7 graphically shows the relationship between Ec′ and SNR with SNR changing. As apparent from FIG. 7 , the value of Ec′ increases rapidly with the increase in SNR. With optical disc devices, the shape of the light spot and the SNR of the optoelectronic transducer change depending on the types of devices or in response to, for example, ambient temperatures. For a storage device such as a hard disc drive in which the disc medium is unchangeably installed, it is the best method to adjust the recording condition in such a manner that SbER (or bit error rate) with respect to the hard disc drive of interest is minimized. However, for a storage system such as an optical disc device in which the disc medium is interchangeable, it is not satisfactory to minimize SbER (or bit error rate) with respect only to that particular device. From the viewpoint of the required performance (2) mentioned above, this method still leaves room for improvement in pursuing the optimal method for obtaining recording condition for highdensity recording.
Further, the reason why this method leaves room for improvement from the viewpoint of the required performance (1) will also be described. The bit arrays used for evaluating Tsfp(3s, 2m) are as described above. On the other hand, the following bit arrays for evaluation are also used for calculating SbER, as described in the patent document 4.
TABLE 2
↓ interested edge
leftshifted bit array
1 1 1 0 0 1 1 1 0 0 1 1 1
correct bit array
1 1 1 0 0 0 1 1 0 0 1 1 1
rightshifted bit array
1 1 1 0 0 0 0 1 1 0 0 1 1
Table 2 given above corresponds to the case where the interested 2T mark is immediately followed by a 2T space. With respect to the leftshifted bit array, the Hamming distance and the Euclidean distance, from the correct bit array are 1 and 14, respectively, just as described above, whereas with respect to the rightshifted bit array, the Hamming distance and the Euclidean distance, from the correct bit array are 3 and 12, respectively. The Hamming distance in this case differs from that in the previous case. From the viewpoint of the required performance (1), it is expected that the correlation between the evaluation index for recording adjustment and the index SbER (or bit error rate) for evaluating the quality of reproduced signals, is sufficiently strong. Therefore, it is necessary that the index representing the evaluation bit array for recording adjustment is roughly the same as the index for evaluating the quality of reproduced signals. The patent document 5 does not disclose any measure for solving the problem that with respect to the evaluation index using the target signal corresponding to the bit array in which the interested edge is shifted to the left or to the right, there arises, as in this instance, plural combinations of (Hamming distance 1 for left edge shift and Hamming distance 2 for right edge shift) and (Hamming distance 1 for left edge shift and Hamming distance 3 for right edge shift). From this point of view, too, it can be said that this method leaves room for improvement.
(4) Method According to Combination of Conventional Techniques
The nonpatent document 2 discloses a technique wherein, on the basis of the technique disclosed in the patent document 1, edge shifts are taken into consideration and adjustment is made in such a manner that the average of distributions of Euclidean distance differences becomes equal to the ideal Euclidean distance. An analogy of a method is easy for making such adjustment that the average of distributions becomes equal to the ideal Euclidean distance, by applying this technique disclosed in the nonpatent document 2 to the evaluation bit arrays shown in “FIG. 3 of the patent document 5”. As shown in FIG. 5 , however, the increase in recording density causes the average of the respective distributions to deviate in the direction in which the average tends to be smaller as compared with the ideal Euclidean distance. In like manner, the average of the respective distributions change in response also to SNRs. FIG. 8 shows the result of the experiment made to ascertain this phenomenon. This experimental result was obtained in reproducing the recorded signals while changing the reproducing (or read) power for the layer L0 of the above described test disc having three layers. In FIG. 8 , the horizontal axis is graduated in reproducing (or read) power with 1.2 mW equated to 100%. The amplitude of the reproduced signal is proportional to the reproducing (or read) power, but the noise (amplifier noise) inherent to the photodetector is constant. In this experiment, the SNR of the reproduced signal is varied by changing the reproducing (or read) power. It is understood from FIG. 8 that the average of the respective distributions is smaller than the Euclidean distance (=1) and that the average decreases with the decrease in the reproducing (or read) power. It is apparent with this method, too, that the difference in SNR depending on the states of drive devices affects the index used to adjust recording.
(5) Method for Minimizing SbER
As shown in FIG. 4 , SbER develops a strong correlation with bit error rate in the experiment regarding recording density of 33 GB per disc surface. Accordingly, a candidate method is to select the condition for obtaining the minimum SbER after having performed write/read operations under all the combinations of possible recording conditions without using the evaluation index for write adjustment. However, it is substantially impossible to search for the condition for obtaining the minimum SbER while randomly changing the recording conditions, in such a case as of an optical disc medium in which the size of the area (test write area) for write adjustment is limited. For it is impossible to obtain the information on the direction for making the edge of the recorded mark approach the ideal state. Except for the conventional techniques as described above which can quantitatively determine the deviations from the target value independently in response to the respective parameters of recording pulses, no method can serve to perform test write applicable to optical disc devices. Further, even in the case where performance improvement is sought while repeating the test fabrication of discs, it is desirable to complete the adjustment of recording condition in a short period of time. In this sense, too, the invention of a novel index and a novel adjusting method has long been waited for according to which the required performances (1) and (2) mentioned above are satisfied and the above mentioned quantitative determination of deviations can be adjusted independently in response to the parameters for recording.
Using a write strategy for conventional BDs, a problem to be solved in order to obtain a preferable recording quality of a high density optical disc having a recording capacity of 30 GB or greater will be described from a point view of a recording pulse.
FIG. 44 shows results of writing random data on an optical disc medium by changing the duration TLP of the last pulse in a pulse string corresponding to each 4T mark, and measuring values of an edge shift of the end edge of each 4T mark based on the readout signal waveforms of the written data. The N−1 write strategy for conventional BDs is used as a write strategy, and the adjustment unit for TLP is set at 1/16 of the channel bit period. TLP is not classified based on the length of succeeding space, and the same value is specified to recording pulses of every 4T mark. An edge shift of VSEAT disclosed in Patent Literature 2 is used as the edge shift. The condition for this experiment is the same as those in the above described experiment using the trial disc.
According to FIG. 44 , in the case of using the write strategy for conventional BDs, if the end position of the last pulse is changed by one unit ( 1/16 of channel bit period), the edge shift is changed by about 16 points. It is preferred that the value of an edge shift should be set within a range of ±5% in order to obtain a preferable recording quality, but it is difficult to always set the value of the edge shift within the range of ±5% at the change rate of the edge shift using the conventional write strategy for BDs.
When using a recording pulse waveform of conventional BDs, an end edge of a recorded mark has a greater change rate of an edge shift, and this is because the parameter regarding formation of the end edge of the recorded mark includes the duration TLP of the last pulse alone, thus if the end position of the last pulse is changed, the duration of the last pulse is also changed. Specifically, since the duration of the last pulse is changed, thermal energy injected into the end edge formed portion of the recorded mark is significantly changed, so that the change rate of the edge shift becomes greater.
A high density optical disc having a recording capacity of 30 GB or greater has a smaller bit length of data written on a disc than that of a conventional BD, so that data is likely to be subjected by thermal influences of writing preceding and succeeding marks. Hence, it is required to control an edge position of a recorded mark more accurately in recording pulses used for a high density optical disc than in those used for conventional BDs.
To solve the above described problem, it can be considered that the adjustment unit of a recording pulse be set at 1/32 of the channel bit period, which is smaller than the unit at 1/16 of the channel bit period, but in this case, data size of the pulse parameters to be stored in control information area of the disc becomes increased, thus it is preferred to accurately control an end edge position of a recorded mark by using the adjustment unit for a recoding pulse set at 1/16 of the channel bit period.
As described above, regarding the adjustment of recording condition corresponding to the highdensity recording condition with recording capacity of more than 30 GB per disc surface in case of a BD system, there was a problem with conventional techniques that the performance of adjustment is not sufficiently compatible with the guarantee of interchangeability of recording media. The objects of this invention are to provide a novel index and a novel method for the adjustment of recording that can solve this problem and to provide an optical disc device using them. The problem to be solved by the present invention is also to provide a recording method capable of controlling an edge position of a mark with high accuracy even under a condition in which the recording pulse adjustment unit is set at 1/16 of the channel bit period.
According to this invention, since the achievement of a large capacity of more than 30 GB is aimed for, it is assumed in the following description of this specification that the minimum run length of the modulation code is 2T. Further, as described above, since the experimental results indicate that the SbER used in the case where up to two 2T continue, coincides well with the bit error rate, it is assumed that SbER is used as the index for evaluating the quality of reproduced signal when the evaluation index for record adjustment according to this invention is discussed. The index for probabilistically evaluating the quality of reproduced signal on the basis of the Euclidean distance between the target signal and the reproduced signal, or the index for directly evaluating bit error rates can yield a good result if the recording condition is adjusted according to this invention.
The summary of the above described requirements is as follows.
[Requirement 1] Compatibility in Reproduction of Data Recorded on the Basis of Result of Adjustment
The evaluation index and the method of adjustment to be employed must not depend on the variation of SNR, but must concentrate on a fixed adjusting target.
[Requirement 2] Quality of Data Recorded on the Basis of Result of Adjustment
In order to secure that SbER is sufficiently small, the evaluation bit array covering at least up to two continuous 2T patterns must coincide exactly or substantially with the evaluation bit array for SbER.
[Requirement 3] Completion of Record Adjustment in a Short Period of Time
The evaluation index and the method of adjustment, capable of independent evaluation, must be provided in accordance with the conditions of recording pulses or the respective parameters of adaptive recording pulses.
[Requirement 4] Highly Accurate Control of End Edge Position of Mark
An end edge position of a mark is required to be controlled with high accuracy even under the condition in which the recording pulse adjustment unit is set at 1/16 of the channel bit period.
According to the concept of this invention, evaluation is performed by separating the component corresponding to the shift of the interested edge from the component depending on SNR in accordance with the difference between the Euclidean distances from the reproduced signal to the two target signals. In order to facilitate the understanding of this invention, the definition of the evaluation index that can solve these problems will first be given, and then the fact will be described that the evaluation index can solve the problems.
In the following description of the specification, W denotes a reproduced signal, T a target signal in the form of a binary bit array obtained from the reproduced signal, L a target signal in the form of a binary bit array in which the interested bit is shifted by a single bit to the left and which satisfies the condition of restriction on run length, and R a target signal in the form of a binary bit array in which the interested bit is shifted by a single bit to the right and which satisfies the condition of restriction on run length. The Euclidean distances between W and T, between W and R, etc. are denoted by ED(W, T), ED(W, R), etc. The evaluation value for the error caused when the interested edge is shifted leftward is represented by xL, and the evaluation value for the error caused when the interested edge is shifted rightward is represented by xR. These evaluation values are called “equivalent edge shifts” and defined with the following expressions (D1) and (D2).
x
L
=
1
2
(
1

ED
(
L
,
W
)

ED
(
T
,
W
)
ED
(
T
,
L
)
)
(
D
1
)
x
R
=
1
2
(
1

ED
(
R
,
W
)

ED
(
T
,
W
)
ED
(
T
,
R
)
)
(
D
2
)
The amount of edge shift of the interested edge is called “extended edge shift D” and defined with the following expression (D3).
D
=
xR

xL
2
(
D3
)
The amount of compensation equivalent to the error probability for the interested edge is called “SNR factor S” and defined with the following expression (D4).
S
=
xR
+
xL
2
(
D4
)
Regarding the interested edge and the group of edges recorded with the same record condition, that is, with a recording pulse that creates a recorded mark having a mark length equal to the space length of a space immediately anterior (or posterior) to the mark, the amount of edge shift used for record adjustment, which is interpreted as the statistical average Δ of extended edge shifts Ds, is defined with the following expression.
Δ = ∑ n = 1 N D n ( D5 )
where N denotes the total number of edges subjected to measurement, and D n the extended edge shift for the nth edge.
Further, the amount of jitter denoted by σ, equivalent to the error probability for the interested edge, is defined with the following expression (D6).
σ = 1 N ( ∑ n = 1 N D n 2 + ∑ n = 1 N S n 2 ) ( D6 )
where S n is the SNR factor for the nth edge.
The quantities, i.e. the “evaluation indices” used in this invention, defined above with the expressions (D1) to (D6) are each called LSEAT (runlengthLimited Sequence Error for Adaptive Target) index. Specifically, Δ defined with the expression (D5) is called LSEAT shift, and σ defined with the expression (D6) is called LSEAT jitter.
According to the adjusting method for the recording condition employed in this invention, recording/reproducing is performed while changing the conditions for recording pulses, and such a particular pulse condition for recording is selected that both the absolute value of the LSEAT shift and the value of LSEAT jitter, for the interested edge are minimized.
Now description is made to ascertain that the above described Requirements 1 to 3 are satisfied by the adjusting method for recording condition according to this invention. As described in the patent documents 15 mentioned above, when PRML procedure is used, the error margin is expressed by the Euclidean distance. In order to facilitate the descriptions to follow, the quantities dEDL and dEDR, which are the values obtained by normalizing the Euclidean distance differences for the errors caused when the interested edge is shifted to the left and to the right, respectively, with the ideal Euclidean distance are defined with the following expressions (3) and (4).
dEDL = ED ( L , W )  ED ( T , W ) ED ( T , L ) ( 3 ) dEDR = ED ( R , W )  ED ( T , W ) ED ( T , R ) ( 4 )
[Requirement 1] Compatibility in Reproduction of Data Recorded on the Basis of Result of Adjustment
As described above, the evaluation index for record adjustment must be constant independent of the change in SNR. The average value of distributions of respective Euclidean distance differences changes in response to the change in SNR. Since W, T, L and R denote the signal levels at plural time instants t (t=t 0 +1, t 0 +2, t 0 +3, t 0 +4, t 0 +5), let them represent the coordinate points in the multidimensional space. For the sake of simplicity, an example is taken of the error caused when a rightward shift of edge occurred and having the resulted Hamming distance of 1. With the (1, 2, 2, 2, 1) characteristic, it can follow that T (T 1 , T 2 , T 3 , T 4 , T 5 ), W (T 1 +δ 1 , T 2 +δ 2 , T 3 +δ 3 , T 4 +δ 4 , T 5 +δ 5 ) and R (T 1 +1, T 2 +2, T 3 +2, T 4 +2, T 5 +1). Further, when the fivedimensional coordinate system is introduced having its origin at T, the position vectors (=coordinate points) for W and R may be represented as W (δ 1 , δ 2 , δ 3 , δ 4 , δ 5 ) and R (1, 2, 2, 2, 1). FIG. 9A illustrates the relative positions of the coordinate points T, W and R in the plane containing these three points. In FIG. 9A , the xaxis extends in the direction along the line segment TR, and normalization is introduced so as to locate the point R at 1 on the xaxis. Further, it should be noted that since the yaxis is taken perpendicular to the xaxis, the direction of the yaxis does not remain fixed but changes depending on the value of W. The Euclidean distances among W, T and R are such as given by the following expression (5).
ED( T,W )+ED( R,W )≧ED( T,R ) (5)
Namely, the sum of the Euclidean distance between T and W and the Euclidean distance between R and W is not necessarily equal to the Euclidean distance between T and R.
FIG. 9B schematically illustrates the measurement of the edge shift of a physically recorded mark. In FIG. 9B , if the distance measured from the target value (the origin) T to the edge of the recorded mark is given by x, then the distance from the edge of the recorded mark to the target value R, which is equivalent to the position shifted by 1T to the right from the origin, becomes equal to (1−x). And the sum of the distances always becomes equal to 1 (=1T: T equals the width of the detection window). In general, the edge control by means of recording pulses is the control of shift along the time axis, being based on the idea of linear measurement for the edge shift of physically recorded marks.
Accordingly, in the definition of the Euclidean distance (equal to the square of the length of line segment), too, if the projection of the vector TW onto the xaxis is represented by xR, then the projection of the vector RW onto the xaxis becomes equal to (1−xR), the sum of these projected components of the vectors TW and RW being equal to 1. The inner product of the vectors TW and TR can produce xR, which can be calculated by using the Euclidean distances among the coordinate points T, R and W, by the following expression.
xR
=
δ
1
+
2
δ
2
+
2
δ
3
+
2
δ
4
+
δ
5
=
1
2
(
1

{
(
1

δ
1
)
2
+
(
2

δ
2
)
2
+
(
2

δ
3
)
2
+
(
2

δ
4
)
2
+
(
1

δ
5
)
2
}

(
δ
1
2
+
δ
2
2
+
δ
3
2
+
δ
4
2
+
δ
5
2
)
14
)
=
1
2
(
1

ED
(
R
,
W
)

ED
(
T
,
W
)
ED
(
T
,
R
)
)
=
1
2
(
1

dEDR
)
(
6
)
This item obtained by the above expression (6) is what is meant by the equivalent edge shift xR defined with the expression (D2) given above. The calculation of equivalent edge shift in the case where the Hamming distance is 2 or 3 can be likewise performed. The second term in the expression (6) indicates the Euclidean distance difference normalized with the ideal Euclidean distance, as shown in FIG. 5 . The quantity xR is not only the projection of W along the direction of the line segment TR but also the quantity related to the error probability that is one of the PRML indices. As a result of natural extension of concept, the equivalent edge shift xL can be calculated by using the target value L, which is equivalent to the position shifted by 1T to the left from the origin, through the expression (D1) given above.
On the other hand, since the coordinates of W change depending on the value of SNR, the value of the equivalent edge shift changes depending on edges to be measured. However, as described above, since the linear addition of equivalent edge shifts is possible in the direction along the line segment TR, it becomes possible to evaluate, independent of SNR, the edge shift of a recorded mark by calculating the average value of the equivalent edge shifts.
The method for coping with the subject that the average of Euclidean distance differences changes depending on SNR will be described below. As described above, the main factor of this phenomenon seems to be ascribed to the fact that the frequency characteristic of the filter implemented by an automatic equalizer is restricted by the Sampling Theorem. Accordingly, when a specific edge is considered, the variations of the average caused as the specified edge shifts to the left and to the right, respectively, are equal to each other. This can be surmised from the fact that the variations of the average of distributions can be classified in terms of the continuous 2T count, i.e. Hamming distances, as shown in FIG. 5 . Let the averages of the normalized Euclidean distance differences dEDL and dEDR be denoted respectively by M L and M R , the respective deviations thereof from the ideal Euclidean distance by dM, and the amount of shift to be measured by Δ 2 . Then, the following equations (7) and (8) hold.
M L =1−Δ+ dM (7)
M R =1+Δ+ dM (8)
On the other hand, according to the technique disclosed in the patent document 2, which uses VSEAT, the normalized sequence errors are calculated depending solely on the edge shifts (Hamming distance 1), plus or minus signs is given the normalized sequence errors with respect to the left or right edge shift, and the arithmetic average of the signed sequence errors is calculated. For example, it is quite natural that the equivalent edge shift to the right is given a plus (+) sign and the equivalent edge shift to the left is given a minus (−) sign. On the basis of this convention of sign allocation, the equivalent edge shifts are calculated with respect to the leftward and rightward shifts of the interested edge. If the arithmetic average of the signed equivalent edge shifts whose signs are given with respect to the edge shift to the left or right, is used as an evaluation value, then the variation dM of the average of distributions of Euclidean distance differences depending on SNR can be offset.
M
R

M
L
2
=
Δ
2
(
9
)
In like manner, it is understood that the extended edge shift D, which is defined with the expression (D3) as a measured value for an interested edge, is the evaluating value for edge shift free of the influence depending on SNR. The LSEAT edge shift Δ defined with the expression (D5) is statistically equivalent to the difference Δ 2 of the respective distribution averages, the Δ 2 being defined with the above expression (9).
FIGS. 10A and 10B diagrammatically show the left and right equivalent edge shifts xL and xR. In these figures, the coordinates of L, R and W are represented in the sixdimensional space of t (t 0 , t 0 +1, t 0 +2, t 0 +3, t 0 +4, t 0 +5) with T assumed as the origin, in consideration of one time instant by which L and R are staggered from the origin. The edge shift x of a physically recorded mark is given by x={(1−x)+(1+x)}/2 where (1−x) is the distance from the anterior edge of the mark to the point R which is staggered by 1T to the right from the point T, and (1+x) is the distance to the anterior edge of the mark from the point L which is staggered by 1T to the left from the point T. Expression (9) means this and measurement. On the other hand, regarding the Euclidean distances among W, T, L and R, since L and R are staggered in time, the line segment TR and the line segment TL are not in alignment with a single geometrical straight line. The angle θ subtended by the two line segments can be given by the inner product of two vectors. If they are errors each of which corresponds to an edge shift (Hamming distance 1), then cos θ is given by the following expression (10).
cos
θ
=
vector
(
T
,
L
)
·
vector
(
T
,
R
)
vector
(
T
,
L
)
vector
(
T
,
L
)
=
0

2

4

4

2

0
14
×
14
=

12
14
(
10
)
In the above expression, vector (T, L) and vector (T,R) represent the position vectors of L and R target signal, respectively, and the operator “·” indicates “inner product”. If T is the most likely target signal and if L and R are the secondary likely (having highest error probability) target signals, then it is reasonable from the viewpoint of error rate in PRML procedure that the recording condition should be so adjusted as to reduce the extended edge shift D to zero. The fact that the two target signals are not on a single geometrical straight line can be said to be the feature of edge shift measurement according to PRML procedure. When the continuous 2T count is 2 (Hamming distance 1, 2 and 3), the relationship among L, R and cos θ is summarized in FIG. 11 . It is seen from FIG. 11 that if the Hamming distance of L is 1 and the Hamming distance of R is 3, then cos θ>0, indicating the angle between vectors L and R being less than 90 degrees, but that if target signals having the highest error probability are selected as L and R, then the edge shift of the interested edge can be measured on the basis of the average Δ of the extended edge shifts D's or the difference Δ 2 of the averages of L and Rdistributions.
FIGS. 12A to 12D graphically show the relationships between dEDL and dEDR, obtained as a result of simulations. The conditions for the simulations were such that the recording density was of 33 GB per disc surface and marks having predetermined lengths were ideally recorded. In these simulations, SNR was equated to 20 dB. FIGS. 12A to 12D correspond respectively to such front edges of marks as (a) Tspf(2s, 2m), (b) Tspf(2s, 3m), (c) Tspf(3s, 2m) and (d) Tspf(3s, 3m), in each of which data were collected from 1000 edges. In these simulations, the target signals having Hamming distances of (a) (2, 2), (b) (2, 1) and (3, 1), (c) (1, 2) and (1, 3) and (d) (1, 1) were used as the L and R target signals. The dashed straight lines in these figures indicate the relationship dEDL+dEDR=2, i.e. relationship equivalent to the conservation of values measured for the physically recorded mark shown in FIGS. 10A and 10B . It is seen from these figures that the plotted data points are scattered roughly along the dashed straight lines, indicating that they develop correlations between dEDL and dEDR, and that the fluctuations with respect to the left and right edge shifts of reproduced signals due to noise are roughly symmetric. To be precise, as seen from FIGS. 12B and 12C , the data distributions indicate gradients a little different from the gradients of the dashed straight lines when the Hamming distances of L and R are not equal to each other. This is because the probabilities of error occurrence for the left and right shifts are different from each other according to PRML procedure, that is, because the measurement of physically recorded marks is different from the measurement on the basis of error margins according to PRML procedure. The evaluation of edge shifts using VSEAT disclosed in the patent document 2 adopts only the target signals having Hamming distance 1 and therefore can only produce the measurement results in which data distribution gradients are parallel to the dashed straight lines even in the cases of Tsfp(2s, 3m) and Tsfp(3s, 2m) of FIGS. 12B and 12C , respectively. The first advantage according to this invention is to have overcome this point.
Each of FIGS. 13A to 13D graphically shows the relationship between the average of dEDL and dERR and the extended edge shift D, obtained as a result of simulations. The conditions for the simulations were the same as in the simulations shown in FIGS. 12A to 12D . These pictures also correspond to (a) Tspf(2s, 2m), (b) Tspf(2s, 3m), (c) Tspf(3s, 2m) and (d) Tspf(3s, 3m), for each of which data were collected from 1000 edges. In these figures, the distributions of averages of dEDL and dEDR (1) spread widely and differently for the different edge patterns and (2) are all shifted toward values smaller than the ideal Euclidean distance difference (=1). This tendency reflects the results shown in FIG. 5 . In contrast to this, the distributions of extended edge shifts D's do not depend on edge patterns, and (1) the spreads of distributions are almost uniform and (2) the center of the spread of each distribution is located nearly at zero. In these figures, these differences are represented diagramatically by different spread shapes. These two advantages (or effects) obtained with the introduction of extended edge shifts D's are due respectively to (1) the calculation of the shift of reproduced signal as the inner product of equivalent edge shift and vector TR or TL and the ensuing linearization, and (2) the averaging of the left and right equivalent edge shifts with minus and plus signs attached thereto, respectively.
FIG. 14 graphically shows the summary of the advantages obtained according to this invention. FIG. 14 shows the relationship between Ec′ (adopted in the method disclosed in the patent document 5) and SNR, as shown in FIG. 7 , superposed with the relationship between averages Δ (defined with the expression (D5)) of extended edge shifts and SNR. Since the optical disc device handles a portable media, it is required to have a function capable of executing a recording to a plurality of various media in the same recording condition. However, SNR of actually recorded information varies according to operational environment of the optical disc device and condition of machine. In a convention method, as shown in FIG. 14 , the value of Ec′ varies largely depending on the variation of SNR. This means that the difference between an ideal recording signal and an actually recorded recording signal (In FIG. 14 , it is evaluated using edge shift amount as barometer) varies according to an operational condition (SNR) of the optical disc device. On the other hand, in the present invention, the Δ value is almost zero and constant without depending on the variation of SNR. This means that recording was executed in the same condition though the operational condition of the optical disc device varied. As stated above, in the present simulation, random noise are added to a signal corresponding to a case that a recording mark of a predetermined length is ideally recorded. The measurement result that the evaluation value Δ for edge shift is nearly equal to zero under such condition indicates high excellence of the method of this invention from the viewpoint of compatibility in the reproduction of recorded data. This point is the second advantage according to this invention.
[Requirement 2] Quality of Data Recorded on the Basis of Result of Adjustment
As a result of the adjustment of recording condition according to this invention, SbER must be sufficiently small. In order to realize this requisite, it is necessary that dEDL and dEDR are minimized through the adjustment of recording pulses and that the evaluation bit arrays for T, L and R are substantially equivalent to the evaluation bit array for SbER.
The former need will first be described. It has been already described that all of the target signals T, L and R are not on a geometrical straight line since they have different Hamming distances and staggered in time from one another. Accordingly, the absolute value of the equivalent edge shift for the edge shift to the left is different from that for the edge shift to the right. This is the feature of the edge shift measurement according to this invention. Now, in evaluating N edges, let dEDL and dEDR of the nth edge be denoted by dEDL n and dEDR n , and let their average be approximated by 1. The standard deviations σ L and σ R can be represented by the following expressions (11) and (12), respectively.
σ
L
=
1
N
∑
n
=
1
N
(
dEDL
n

1
)
2
(
11
)
σ
R
=
1
N
∑
n
=
1
N
(
dEDR
n

1
)
2
(
12
)
The bit error rate can be evaluated by using the synthetic standard deviation σ LR of them represented by the following expression (13).
σ
LR
=
1
2
N
∑
n
=
1
N
(
dEDL
n

1
)
2
+
(
dEDR
n

1
)
2
=
1
4
N
∑
n
=
1
N
{
(
dEDR
n

1
)

(
dEDL
n

1
)
}
2
+
1
4
N
∑
n
=
1
N
{
(
dEDR
n

1
)
+
(
dEDL
n

1
)
}
2
=
4
N
∑
n
=
1
N
(
1

dEDR
n
2

1

dEDL
n
2
2
)
2
+
4
N
∑
n
=
1
N
(
1

dEDR
n
2
+
1

dEDL
n
2
2
)
2
=
2
1
N
∑
n
=
1
N
(
xR

xL
2
)
2
+
1
N
∑
n
=
1
N
(
xR
+
xL
2
)
2
=
2
1
N
(
∑
n
=
1
N
D
n
2
+
∑
n
=
1
N
S
n
2
)
(
13
)
The righthand side of the above expression (13) becomes equal to double the value of the LSEAT jitter given by the expression (D6). The factor 2 appearing on the righthand side is not essential, but results from the fact that the error margin is ±½T for LSEAT as in the case of jitter measurement by the conventional time interval analyzer, whereas the error margin is 1 (ideal Euclidean distance=1) in the distributions of dEDL and dEDR. If the distributions of dEDL and dEDR are regarded as of Gaussian, the error rates for these distributions obtained by using the error function become equal to each other. It is apparent that the LSEAT jitter represents the synthetic standard deviation obtained by superposing the distributions of the Euclidean distance differences shown in FIG. 5 in such a manner that the average of the distributions occurs at the ideal Euclidean distance (=1). It can therefore be said that the LSEAT jitter is the evaluation index that exhibits a strong correlation with SbER or bit error rate. To be further precise, as seen from the expression (D4), the SNR factor by definition has as its average a value equal to the quantity by which the distribution of Euclidean distance differences is deviated from the ideal value (=1) depending on SNR and recording density. Therefore, the contribution of SNR to the LSEAT jitter defined with the expression (D6) involves the deviation of the average of the distributions of Euclidean distance differences. As described above, the LSEAT jitter according to this invention can be evaluated by separating the component corresponding to the shift of the interested edge (the extended edge shift) from the component depending on SNR (the SNR factor). In this way, two functions can be simultaneously provided: one is the shift adjustment excellent in reproduction compatibility independent of SNRs of different drive devices, and the other is the warranty of rendering SbER and bit error rate to the minimum. Namely, in comparison with those conventional signal evaluation indices for recording adjustment which include the VSEAT disclosed in the patent document 2, the LSEAT indices according to this invention develop correlation with such conventional indices for evaluating the quality of reproduced signal as jitter and SbER, and therefore can be regarded as excellent signal evaluation indices. Thus, according to this invention, the excellent signal evaluation indices that no conventional signal evaluation indices have ever paralleled in performance, can be provided, and this fact is the third advantage attained by this invention. The empirical proof of this point will be given later in reference to experimental results.
The affinity to the evaluation bit arrays used for evaluating the quality of reproduced signal in the calculation of SbER will now be described. The techniques for evaluating reproduced signals disclosed in the patent documents 1, 3 and 4 employ different constitutions but include in common the process of searching/extracting the first most likely evaluation bit array from among binary bit arrays outputted from the PRML decoder. The length M of the evaluation bit array can be generalized to give the equation M=2N−1+2N 2T , by using the constraint length N in the PRML procedure and the number N 2T of continuous 2T patterns contained in the evaluation bit array. Here, N 2T denotes an integer such as 0, 1, 2, . . . . According to the notation described above, N 2T =0, 1, 2 correspond to an edge shift, a 2T shift and a 2T knockon, respectively. Further, when N 2T is 0, 1, 2, 3, 4, 5 and 6, the corresponding Hamming distances are 1, 2, 3, 4, 5, 6 and 7, respectively, and the Hamming distance between the evaluation bit arrays of the patterns A and B is (N 2T +1). The evaluation bit array can be easily enumerated from among 2 M bit arrays through the mechanical operations of extracting the relationship between the first most likely evaluation bit array and the second evaluation bit array corresponding to the target signal that makes minimum the Euclidean distance from the target signal corresponding to the first evaluation bit array.
FIG. 15 is a table listing examples of evaluation bit arrays associated with PR (1, 2, 2, 2, 1) characteristic having a constraint length of 5. Similar bit arrays are disclose in the patent document 4. As is apparent from FIG. 15 , in the case where evaluation bit arrays are searched and extracted from among binary bit arrays outputted from PRML decoder, using PRML procedure having a constraint length of 5, the total of 108 evaluation bit arrays, i.e. 54 pairs including 18 pairs for each of the Hamming distances, are enumerated. In evaluating reproduced signals, these evaluation bit arrays must be searched and extracted simultaneously.
FIG. 16 is a table listing the evaluation bit arrays associated with PR(1, 2, 2, 2, 1) characteristic having a constraint length of 5, shown in FIG. 15 , in such an abridged way that those bit arrays which are equivalent to one another except their two head bits and two tail bits, are grouped together. As seen from FIG. 16 , the 108 evaluation bit arrays associated with Hamming distances 1, 2 and 3 are represented by the main bit arrays having bit lengths of 5, 7 and 9 and the auxiliary bit arrays XX and YY each having a bit length of 2 and attached to the front and rear end of these main bit arrays. Here, the main bit arrays include four bit arrays such as “00011”, “00111”, “11100” and “11000” associated with Hamming distance of 1; four bit arrays such as “0001100”, “0011000”, “1110011” and “1100111” associated with Hamming distance of 2; and four bit arrays such as “000110011”, “001100111”, “111001100” and “110011000” associated with Hamming distance of 3. The auxiliary bit array AA is “00”, “10” or “11” and the auxiliary bit array BB is “00”, “01” or “11”. The intervals corresponding to the bit lengths of the main bit arrays listed here are used as the intervals within which the Euclidean distance between target signal and reproduced signal is calculated. The auxiliary bit arrays are only used to calculate the levels of target signals at the front and rear ends of the main bit arrays and have nothing to do with the calculation of Euclidean distances among target signals. In this sense, the auxiliary bit arrays can be considered to determine the boundary condition to determine the levels of the target signals at their ends.
The main bit arrays can be determined independent of the constraint length in PRML procedure. The reason for this will be described below. If the shortest run length m is set equal to 2T, then in order to show that a single bit is changed due to an edge shift, the shortest length in terms of bit is obtained by multiplying m by 2 and adding 1 to 2m, that is, equal to 2m+1=5 bits. This is what the main bit looks like. In like manner, the generalization using the continuous 2T count N 2T included in each evaluation bit array yields the length of main bit array equal to 2m+1+2N 2T . Thus, the main bit array is meant as the shortest bit array determined depending on the continuous 2T count contained in an evaluation bit array. On the other hand, as described above, the length of bit array needed to calculate the Euclidean distance from reproduced signal is represented, by using the constraint length N in PRML procedure, as 2N−1+2N 2T . The difference between the lengths of both the bit arrays is (2N−1+2N 2T )−(2m+1+2N 2T )=2(N−m−1), which is always an even number. This value is 2(N−3) if the shortest run length m=2.
As described above, if use is made of the main bit array independent of the constraint length N in PRML procedure and the auxiliary bit arrays having length (N−3) and attached at the front and rear ends of the main bit array, then the evaluation bit array can be represented in some ordered manner.
In this way, the ordered representation of each evaluation bit array enables the relationship between the index for evaluating the quality of reproduced signal and this invention to be simplified and also the scale of circuitry used in this invention to be reduced.
In the table of FIG. 16 , evaluation bit arrays are classified into A and B groups in accordance with the disclosure of the patent document 4. It is preferable from the viewpoint of reducing the scale of circuitry to select the first evaluation bit array (i.e. evaluation bit array equivalent to target signal T) from among bit arrays obtained by binarizing reproduced signals and then to generate the second evaluation bit array as secondary likelihood bit array (i.e. evaluation bit array equivalent to target signal L or R) on the basis of first evaluation bit array. Since the Hamming distance between the first and second evaluation bit arrays is previously determined, the second evaluation bit array can be generated by applying exclusive OR (XOR) operation to the first evaluation bit array (T) and the generation bit array having 1's whose number is equal to the Hamming distance. FIG. 17 is a table which summarizes main bit arrays corresponding to Hamming distances 1 to 7. In the column for the main bit arrays of the table, main bit arrays given above are enumerated. In this table, the numbers of the main bit arrays are determined by combining the Hamming distances and the numbers 1 to 4. As shown in the figure, the second main bit arrays can be obtained through XOR operations using specific generation bit arrays peculiar to Hamming distances. Also, the main bit array numbers of the second main bit arrays are enumerated in the table.
As described above, the consideration of main bit arrays leads to the explanation of the affinity between the evaluation bit array for SbER and the evaluation bit array according to the method of this invention.
FIG. 18 is a table listing main bit arrays for edge evaluation according to this invention in the case where the continuous 2T count is equal to or less than 2. When the L and Rtarget generation bit arrays are simultaneously generated to calculate the LSEAT index, the length of each main bit array to be used is greater by 1T (i.e. one bit) than that of the corresponding main bit array shown in the table of FIG. 17 . Thus, the target generation arrays having Hamming distances of 1, 2 and 3 have bit lengths of 6, 8 and 10, respectively. In FIG. 18 , just as in FIG. 17 , listed are the main bit arrays to be included in the bit arrays obtained through the binarization of reproduced signals and the L and Rtarget generation bit arrays for generating the L and Rtarget signals by performing XOR operations on themselves and the main bit arrays. The total number of the listed main bit arrays is 12, and the underlined bit of each main bit array is the interested edge. The rule adopted here is to select as the L and Rtarget generation bit arrays the main bit arrays whose interested edges are shifted by a single bit to the left and right, respectively, which satisfy the run length restriction, and whose Hamming distances are minimized (i.e. the number of inverted bits is minimized). Also, in the table in FIG. 18 , the recorded mark is indicated by “1” and the space is indicated by “0”. Even when the amount of light reflected from the recorded mark is less than that of light reflected from the space, that is, in the case of the socalled HightoLow type recording medium being used, the PR(1, 2, 2, 2, 1) characteristic can be maintained if the “1” and “0” of the main bit array are inverted so as to cause the recorded mark and the space to be denoted by “0” and “1”, respectively. Alternatively, if all the impulse responses are inverted with PR(−1, −2, −2, −2, −1) characteristic employed, the table shown in FIG. 18 can be used as it is. In the description of this invention given hereafter, the recorded mark is represented by “1” and the space by “0” unless otherwise specified.
Description is made below of the relationship between the main bit arrays shown in FIG. 17 for calculating SbER and the main bit arrays shown in FIG. 18 for evaluating LSEAT. FIGS. 19A and 19B illustrate the comparison between both types of the main bit arrays in the case where N 2T =0, that is, Hamming distance is 1. This comparison was made to perform the evaluation of the anterior edge of a recorded mark having a length of 3T or longer. The time instants at which the evaluation is made are included with respect to SbER while the types of edges are included with respect to LSEAT. As shown in these figures, each main bit array includes only one edge. With respect to both SbER and LSEAT, evaluation for two Hamming distances are made per edge. The main bit arrays for SbER coincides with the main bit arrays for LSeat. Namely, the evaluation bit arrays for SbER and LSEAT, including their auxiliary bit arrays, coincide with each other. In these figures, only the evaluation of the anterior edge is performed, but if “1” and “0” are inverted, the evaluation for the posterior edge can be likewise performed. It is evident in that case, too, that the evaluation bit arrays for SbER and LSEAT remain coincident with each other.
FIG. 20 illustrates the comparison between both the main bit arrays in the case where N 2T =1, that is, Hamming distance is 2. Each of the main bit arrays includes two edges. With respect to both SbER and LSEAT, evaluation for two Hamming distances are made per edge. The main bit array for SbER coincides with the main bit array for LSEAT. Regarding the transition of evaluation with time, as indicated with arrows in the figure, it is apparent that evaluation is performed for L, L, R and R edges in this order named in case of SbER and for L, R, L and R edges in this order named in case of LSEAT. The evaluation bit arrays for SbER and LSEAT would coincide with each other also in the case of bit array patterns in which “1” and “0” are inverted.
FIG. 21 illustrates the comparison between the main bit arrays for both SbER and LSEAT in the case where N 2T =2, that is, Hamming distance is 3. Each of the main bit arrays includes three edges. As in the above example, with respect to both SbER and LSEAT, evaluation for two Hamming distances are made per edge. The main bit arrays for SbER coincide with the main bit arrays for LSeat. Regarding the transition of evaluation with time, as indicated with arrows in the figure, it is apparent that evaluation is performed for L, L, L, R, R and R edges in this order named in case of SbER and for L, R, L, R, L and R edges in this order named in case of LSEAT. The evaluation bit arrays for SbER and LSEAT would coincide with each other also in the case of bit array patterns in which “1” and “0” are inverted.
It has been revealed from the foregoing discussion that in the case where N 2T is equal to or less than 2, the evaluation bit arrays for calculating SbER coincide with the main bit arrays for evaluation shown in FIG. 18 . In like manner, with respect also to the case where N 2T is equal to or greater than 3, the evaluation bit arrays for SbER and LSEAT can be likewise made coincident with each other if the main bit arrays for calculating LSEAT, whose maximum Hamming distance is the same as the maximum Hamming distance of the main bit arrays for calculating SbER, satisfy the run length restriction and if the main bit arrays whose Hamming distances are minimum are selected as the main bit arrays for the L and R target signals. A concrete example in which N 2T is equal to 3 will be described later. The basic concept of LSEAT is to evaluate the edge shift on the basis of the difference of averages of distributions of the target signals whose edges to be evaluated are shifted to the left and right, by focusing attention to symmetricity, when the average of distributions of Euclidean distance differences differs from the difference of Hamming distances as shown in FIG. 5 . According to this concept, the edge shift is evaluated on the basis of the procedure (expressions D1 to D6) in which the extended edge shift is evaluated at respective time instants, or the procedure (expressions 7 to 13) in which the average of distributions of independently calculated Euclidean distance differences is evaluated. Further, not only main bit arrays for evaluation shown in FIG. 18 but also various variations including a case of N 2T =3, can be used.
As described above, if the main bit arrays for evaluation shown in FIG. 18 are evaluated on the basis of the equivalent edge shifts, the LSEAT can be provided as an evaluation index with improved correlation with SbER or index of common concept, also from the viewpoint of the affinity of the evaluation main bit arrays. This point is the fourth improvement achieved by this invention.
[Requirement 3] Achievement of Record Adjustment in a Short Period of Time
The evaluation index and the method of adjustment, capable of independent evaluation must be provided in accordance with the condition for recording pulses or the respective parameters of adaptive recording pulses. Optical disc devices in general need to handle not only optical discs of a particular standard but also optical discs for highdensity recording such as CD, DVD, BD or other optical discs fabricated based on BDs. Adaptive recording pulses may vary in accordance with the standards for these different discs. It is also preferable that the proper index for evaluating such recording adjustments as, for example, the edge shift along the time axis measured by a time interval analyzer, the jitter, the VSEAT or the LSEAT according to this invention, should be used depending on cases to be treated. In order to realize such a flexible index, a parameter table for recording adjustment has only to be prepared first. Then, allround measures can be provided by installing in the previous stage a circuit for calculating such evaluation indices as the edge shifts of reproduced signals and the SNR factors.
FIG. 22 shows in block diagram the configuration of such a circuit for adjusting recording condition. In FIG. 22 , the readout signal 51 , which has been read out from an optical disc medium and passed through an analog filter not shown in the figure, is converted to a digital signal having 6 to 8 bits by an A/D converter 21 , equalized by an automatic equalizer 22 , and binarized by a PRML decoder 23 to be output as a binary bit array 52 . A circuit 30 for evaluating the quality of the signal for adjusting the write condition comprises edge evaluators 40 , 41 and 42 ; a selector 60 ; a write pulse evaluation table 35 ; and a timing adjuster 36 . The edge evaluator 40 performs the evaluation of the edge shift along time axis with respect to each edge in a CD/DVD system; the edge evaluator 41 performs the evaluation of VSEAT for BD; and the edge evaluator 42 performs the evaluation of LSEAT for highdensity BD. Each edge evaluator calculates the amount of edge shift, the extended edge shift or the SNR factor with respect to each edge. The selector 60 selects the outputs of the edge evaluators depending on the kinds of discs subjected to write/read operation. The write pulse evaluation table 35 synchronizes the binarized bit array 52 with the edge evaluation index outputted from the edge evaluator, performs the classification of bit patterns in accordance with adaptive recording pulses, allocates the bit patterns into, for example, a 4×4 table, and calculates the average or standard deviation per table component. A CPU 140 refers to the result obtained by the write pulse evaluation table 35 and processes the adjustment of the respective parameters of the adaptive recording pulses. With this configuration described above, the parallel adjustments of parameters of the adaptive recording pulses becomes possible for different kinds of optical disc media. Accordingly, the adjustment of condition for write pulses can be achieved in a short period of time and in a limited test write area, in comparison with the method that uses a single evaluation index for readout signal.
[Requirement 4] Highly Accurate Control of End Edge Position of Mark
Descriptions will now be provided on the solution of controlling an end edge position of a recorded mark with high accuracy in a high density optical disc having a recording capacity of 30 GB or greater, even under the condition in which the recording pulse adjustment unit is set at 1/16 of the channel bit period.
In recording pulses of conventional BDs, an ending position of a last pulse is changed by changing a duration of the last pulse, thus amount of thermal energy injection into an end edge formed portion of the mark is changed, which causes a problem of increase in change rate of an end edge position of the mark due to the change of the ending position of the last pulse.
To address this problem, the present invention employs a parameter of dTLP for specifying a starting position of the last pulse. FIG. 45 illustrates recording pulse waveforms in which the dTLP parameter of the present invention is applied to the write strategy for a conventional BD. The dTLP parameter is defined as illustrated in FIG. 45 , on the basis of the NRZI channel bit signal of written data.
The position of the last pulse can be changed without changing the duration of the last pulse by changing dTLP while fixing TLP. In this case, the change in the amount of thermal energy injection into the end edge formed portion of the recorded mark is small, so as to keep small the change rate of the end edge position of the mark relative to the change of the ending position of the last pulse. It can be understood that the reason why this method can change the end edge position of the recorded mark is because a size of a gap immediately before the last pulse is changed, which causes a change of a maximum endpoint temperature at the mark endedge formed portion.
FIG. 46 illustrates a parameter table of the last pulse starting position dTLP that is employed in the present invention. The values of dTLP are classified and specified into 3×4 combination patterns in total, which are constituted by three recorded mark lengths of 3T, 4T and 5T or greater, and four succeeding space lengths of 2T, 3T, 4T and 5T or greater.
Descriptions will now be provided on the advantageous effects by employing the parameter dTLP for specifying the last pulse starting position. FIG. 47 shows measurement results in which, while changing the last pulse ending position corresponding to the 4T mark as similar to the experiment of FIG. 44 , an edge shift of an end edge of each 4T mark is measured. In this experiment, the last pulse ending position is changed by using two methods, and results by the methods are plotted, respectively. One method depends on recording pulses of a conventional BD, and changes the last pulse ending position by changing the last pulse duration TLP alone while fixing the last pulse starting position. The other method depends on the recording pulses of the present invention, and changes the last pulse ending position by changing the last pulse starting position dTLP while fixing the last pulse duration TLP. The adjustment unit for each last pulse ending position in both the methods is set at 1/16 of the channel bit period. The experimental conditions such as the writeonce optical disc samples used and the recording/reproducing condition are the same as in the above described experiment. The duration TLP of a last pulse of every 4T mark recording pulse string is changed by using the same value regardless of the succeeding space length thereof. At this time, the adjustment unit for each TLP value is set at 1/16 of the channel bit period in accordance with a conventional BD.
According to FIG. 47 , in the case of changing the last pulse ending position by using the recording pulses of the conventional BD (TLP control), the edge shift of the end edge of the recorded mark is changed by about 16 points when changing the last pulse ending position by one unit; contrary to this, in the case of using the recording pulses of the present invention (dTLP control), the edge shift is changed by 6 points. Accordingly, it is supposed that the value of the edge shift can always be set within the range of ±5%, which indicates the preferable recording quality as described above, if dTLP of the present invention is used.
In this way, it has become possible to provide the evaluation index for adjustlong the recording condition, the adjusting method for recording condition and the optical disc device using the evaluation index and the adjusting method, which, regarding the adjustment of write condition corresponding to highdensity recording with recording capacity of more than 30 GB per disc surface based on BD systems, can solve the above described problems peculiar to the conventional techniques and which can (1) enjoy the high compatibility of the reproduction of data recorded in accordance with the result of the adjustment, (2) guarantee that the quality of the data recorded in accordance with the result of the adjustment is sufficiently good when it is measured in terms of the evaluation index for reproduced signal such as SbER, and (3) adjust the condition of the adaptive recording pulse in a short period of time.
The gist of this invention is the provision of the adjusting method for recording condition and the optical disc device using this method, wherein when reproduced signals are evaluated by using the target signals having more than three Hamming distances (corresponding to N 2T =1, 2 and 3) with respect to an optical disc using recorded marks whose shortest run length is equal to 2T as in BD, the quality of the interested edge is evaluated in accordance with the method for evaluating extended edge shifts at respective time instants or the method for evaluating the average of distributions of independently calculated Euclidean distance differences; and the recording condition is adjusted on the basis of the result of the evaluation.
Effect of the Invention
As described above, an optical disc device capable of achieving highdensity recording of 30 GB or higher in BD can be realized by using the adjusting method for recording condition which uses the LSEAT according to this invention as evaluating index.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 shows in block diagram the configuration of the circuit for evaluating the reproduced signal, the circuit being used to realize an optical disc device according to this invention;
FIG. 2 graphically shows the experimental results illustrating the relationship between recording power and bit error counts, measured by using a 3layer writeonce optical disc fabricated for test purpose;
FIG. 3 graphically shows the results of simulations illustrating the relationship between SNR and SbER;
FIG. 4 graphically shows the experimenter result illustrating the relationship between bit error rate and SbER;
FIG. 5 graphically shows relations between recording capacities and distributions of Euclidian distance differences.
FIG. 6 graphically shows distributions of Euclidian distance differences obtained through simulation, which are used for evaluating Tsfp(3s, 2m).
FIG. 7 graphically shows the relationship between SNR and Ec′;
FIG. 8 graphically shows the relationship between reproduction power and deviation of distribution center;
FIGS. 9A and 9B diagramatically show an example of equivalent edge shift;
FIGS. 10A and 10B schematically show another example of equivalent edge shift;
FIG. 11 shows the relationship between Hamming distance and cos θ;
FIGS. 12A to 12D show the correlations between dEDL and dEDR;
FIGS. 13A to 13D show the relationships between the averages of dEDR's and dEDR's and the extended edge shifts;
FIG. 14 graphically shows the relationship between SNR and Ec′, and the average A of extended edge shifts;
FIG. 15 is a table listing evaluation bit arrays associated with PR(1, 2, 2, 2, 1) characteristic;
FIG. 16 is a table listing evaluation bit arrays associated with PR(1, 2, 2, 2, 1) characteristic, the bit arrays being extracted according to similarities;
FIG. 17 is a table related to the main bit arrays and the generation of the second main bit arrays;
FIG. 18 is a table listing evaluation main bit arrays (N 2T max =2);
FIGS. 19A and 19B show the comparisons between evaluation main bit arrays for SbER and LSEAT;
FIG. 20 shows comparisons between evaluation main bit arrays for SbER and LSEAT;
FIG. 21 shows additional comparisons between evaluation main bit arrays for SbER and LSEAT;
FIG. 22 shows in block diagram an evaluation circuit;
FIG. 23 is a table listing evaluation main bit arrays (N 2T max =3);
FIG. 24 is a table as another embodiment, listing evaluation main bit arrays (N N 2T max =2);
FIG. 25 is a table as another embodiment, listing evaluation main bit arrays (N N 2T max =3);
FIGS. 26A to 26C show correspondences between evaluation bit arrays and write pulse tables;
FIGS. 27A to 27C show additional correspondences between evaluation bit arrays and write pulse tables;
FIGS. 28A and 28B illustrate the evaluations of edge shifts by using LSEAT;
FIGS. 29A and 29B illustrate another example of the evaluations of edge shifts by using LSEAT;
FIG. 30 shows LSEAT distributions and SAM distributions;
FIGS. 31A and 31B graphically show the relationships between read powers and LSEAT evaluation indices;
FIG. 32 shows in block diagram the structure of an automatic equalizer of symmetric type;
FIG. 33 graphically shows experimental results regarding recording adjustment using LSEAT;
FIG. 34 graphically shows additional experimental results regarding recording adjustment using LSEAT;
FIG. 35 graphically shows yet additional experimental results regarding recording adjustment using LSEAT;
FIG. 36 graphically shows still additional experimental results regarding recording adjustment using LSEAT;
FIG. 37 graphically shows power margin after write adjustment;
FIG. 38 shows the relationship between bit error rates and LSEAT jitter and the relationship between bit error rates and VSEAT jitter;
FIG. 39 diagrammatically shows how write pulses are adjusted;
FIG. 40 is the flow chart illustrating a method for write adjustment according to this invention;
FIG. 41 graphically shows how focus is adjusted in accordance with the evaluation method according to this invention;
FIG. 42 schematically shows the structure of an optical disc device.
FIG. 43 illustrates recording pulse waveforms of a conventional BD.
FIG. 44 is a diagram showing the relation between a last pulse ending position and an edge shift at an end edge of a recorded mark in the case of using recording pluses of the conventional BD.
FIG. 45 illustrates recording pulse waveforms of the present invention.
FIG. 46 is a parameter table of a last pulse starting position dTLP of the present invention.
FIG. 47 is a diagram showing the relation between a last pulse ending position and an edge shift at an end edge of a recorded mark in the case of using recording pulses of the present invention.
FIG. 48 is a flow chart of illustrating the recording pulse adjustment method of the present invention.
DETAILED DESCRIPTION OF EMBODIMENTS
Embodiments of the present invention will be described below.
Embodiment 1
The adjusting method for recording condition and the optical disc device according to this invention will now be described by way of embodiment in reference to the attached drawings.
FIG. 23 is another embodiment of the table listing main bit arrays for edge evaluation according to this invention. The listed main bit arrays are for the case where N 2T is equal to 3. There are 20 main bit arrays listed in total. In each main bit array, the underlined bit corresponds to an interested edge. The main bit arrays No. 1 to 12 are the same as those listed in the table shown in FIG. 18 . The main bit arrays No. 13 to 20 correspond to the case where the continuous 2T count is 3. As described above, there is only one evaluation bit array that is the secondary likelihood bit array, among main bit arrays listed in FIG. 15 or FIG. 16 . Accordingly, even in case of such a binarized bit array with N 2T =3 as “0000011001100”, the quality of reproduced signal is evaluated by regarding the bit array “0000110011000” having Hamming distance=3 as the secondary likelihood evaluation bit array. On the other hand, this is not evaluated for the evaluation main bit arrays listed in FIG. 15 . In the case where such a bit array must be independently evaluated depending on recording density and disc medium, that is, where when the continuous 2T counts are 2 and 3, the difference between the edge shifts of recorded 2T marks cannot be ignored, the evaluation main bit array shown in FIG. 23 must be used though the scale of the circuit used in that case will be increased. Moreover, by using evaluation main bit arrays shown in FIG. 23 , the main bit array (No. 15) with the mark having a length of 3T or longer preceding Tsfp(2s, 2m) and the main bit array (N. 17) with the mark having a length of 2T preceding Tsfp(2s, 2m) can be separately evaluated. Further, in the case where adaptive recording pulses are used for actual recording in accordance with not only the length of a space preceding a record mark but also the length of a mark appearing still further ahead, the evaluation main bit arrays shown in FIG. 23 can produce information on the recording adjustment condition that exhibits one to one correspondence with the recording pulse table. Regarding the continuous 2T count (N 2T ) included in each evaluation main bit array, suitable one may be employed in consideration of such a circumstance as described above. The above described evaluation main bit array exhibits one to one correspondence with the evaluation bit array for calculating SbER as in the case of the main bit array with N 2T =2. The description of the case where N 2T is 4 or greater, will be lengthy and therefore omitted here. However, those who are skilled in the optical disc technology will easily be able to extend their idea to such a case from relationships of FIG. 18 and FIG. 23 .
FIG. 24 is another embodiment of table listing main bit arrays for edge evaluation according to this invention. Here is shown the case where Hamming distances for L and R target signals are set equal to each other when N 2T =2. There are 12 main bit arrays listed in total. In each main bit array, the underlined bit corresponds to an interested edge. FIG. 24 differs from FIG. 18 in L and Rtarget generation bit array and Hamming distance. If the evaluation main bit arrays shown in FIG. 18 are used, their onetoone correspondence with the bit arrays for evaluating SbER cannot be maintained, but the deviation of distribution relative to SNR can be offset in principle. Accordingly, the dependence on SNR (constant in principle) better than the dependence on SNR shown in FIG. 14 can be obtained. These evaluation main bit arrays may be employed if the change in SNR due to the change in drive device and recording medium is regarded as the first priority item.
FIG. 25 is still another embodiment of table listing main bit arrays for edge evaluation according to this invention. Here is shown the case where Hamming distances for L and R target signals are set equal to each other when N 2T =3. The feature of the table shown in FIG. 25 and the associated evaluation performance are the same as those as described with respect to FIG. 24 . In this case, too, a good dependence on SNR (constant in principle) can be obtained.
FIGS. 26A to 26C and FIGS. 27A to 27C show correspondences between evaluation main bit arrays and recording pulse tables, as embodiments of this invention. FIGS. 26A to 26C shows the same main bit arrays as in FIG. 18 and the recording pulse tables showing anterior edges Tsfp and posterior edges Telp with 4×4 different combinations of marks and spaces (referred to also as 4×4 type pulse table), giving associated main bit array numbers. As seen in FIGS. 26A to 26C , the main bit array of No. 9 may be used corresponding to the result of evaluation of Tsfp (2s, 2m). If a circuit for recording adjustment in a drive device is constructed in reference to this table, LSEAT can be used to adjust the parameters of the recording pulses as defined in the 4×4 type pulse table. FIGS. 27A to 27C show correspondences between evaluation main bit arrays and the associated Hamming distances of the L and Rtarget signals in such a case. In this way, the result of edge evaluation using LSEAT can be easily made coincident with the parameter table for the recording pulses. The main bit arrays shown in FIGS. 23 to 25 can be likewise evaluated in accordance with the parameter table for the recording pulses.
FIGS. 28A and 28B and FIGS. 29A and 29B show the simulation results illustrating examples of the edge evaluation using LSEAT in accordance with the above described parameter table for the recording pulses. Here, the condition for simulation is the same as that described above, the recording density is equivalent to 33 GB per disc surface as of a BD, and the PR class is (1, 2, 2, 2, 1).
The simulation in this case was performed when Tsfp(2s, 2m) is +0.2T (shift by 0.2T to the right). FIGS. 28A and 28B correspond to the case where the method for evaluating extended edge shifts at respective time instants (expressions (D1) to (D6)) is used. It is apparent from FIG. 29B that when Tsfp(2s, 2m) is shifted to the right, the corresponding distribution is also shifted to the right. A good recording condition can be obtained by adjusting the parameters of the recording pulse so that respective edge shifts may approach zero. FIGS. 29A and 29B correspond to the case where the method for evaluating the average of distributions of the independently calculated Euclidean distance differences (expressions (7) to (13)) is used. In the simulation, the Hamming distances in FIGS. 29A and 29B are both set equal to 2 for the left and right shifts. As seen in FIG. 29A , when the shift of Tsfp(2s, 2m) is zero, the averages of the L and Rdistributions differs from the ideal Euclidean distance (=1), but both the averages are the same as each other within a tolerable range of error. On the other hand, as seen in FIG. 29B , when the shift of Tsfp(2s, 2m) is not zero, the averages of the L and Rdistributions shift to the opposite directions. Therefore, if the parameters of the recording pulse are adjusted so that the averages of the L and Rdistributions may coincide with each other, a good recording condition can be obtained. In this way, if the Hamming distances of the main bit arrays for evaluating the L and Rtarget signals are set equal to each other, the recording condition can be adjusted independent of SNR by using symmetricity. As described above, main bit arrays having different Hamming distances can also be employed as the main bit arrays for evaluating the L and Rtarget signals.
FIG. 29 shows the comparison between the SAM distribution and the LSEAT distribution, both distributions being obtained through simulation. Here, the condition for simulation is the same as that described above, the recording density is equivalent to 33 GB per disc surface as of a BD, and the PR class is (1, 2, 2, 2, 1).
It is ascertained that the average of SAM distribution approaches zero as SNR decreases whereas the average of LSEAT distribution remains fixed to zero independent of SNR. Since the case where the evaluation main bit arrays have N 2T =3 or greater, is an extended version of the case shown in FIG. 30 , the same result can be obtained.
FIG. 30 graphically shows the experimental results regarding the dependence of LSEAT indices on SNR. These results was obtained by performing read experiments while the read power for the L0 layer of the above mentioned 3layer disc fabricated for test purpose, was being changed. The results correspond to the result shown in FIG. 8 obtained according to the conventional technique. In FIG. 30 , the horizontal axis is graduated in read power with the read power of 1.2 mW normalized as 100%. Since the amplitude of the readout signal is proportional to the read power while the noise in the photodetector (amplifier noise) is constant, then in this experiment the SNR of the readout signal is changed by changing read power. The LSEAT jitter and LSEAT shift were evaluated with the evaluation circuit shown in FIG. 22 , through allocation to the 4×4 type pulse table with respect to the anterior and posterior edges of a recorded mark. FIG. 31A shows the measured values of LSEAT jitter. The increase in the jitter with the decrease in the read power reflects the change in SNR. On the other hand, FIG. 31B shows the result of evaluation of edge shift with respect to Tsfp(2s, 2m). It is understood from FIG. 31B that the value of the LSEAT shift remains constant independent of read power (SNR). This characteristic is the feature of the method according to this invention which makes it possible to evaluate the marginal evaluation based on the basis of the Euclidean distance difference by separating the component of the edge shift and the component depending on SNR from each other on the basis of LSEAT. It has therefore been ascertained that the use of this method enables the adjustment of write condition with high read compatibility independent of the change in SNR caused depending on the difference among drive devices and read/write conditions.
Now, description is made of an automatic equalizer suitable for adjusting the recording condition.
FIG. 32 shows in block diagram the structure of an automatic equalizer of symmetric type according to this invention. As described above, the use of LSEAT makes it possible to stabilize the adjustment of the recording pulses in response to change in SNR. On the other hand, drive devices for actual use encounter (1) asymmetricity in the scanning direction of light spot due mainly to the relative tilt (tangential tilt) angle between disc medium and optical head, and (2) asymmetricity of readout signal along the time axis due to the asymmetricity of the tap coefficients of the automatic equalizer. The distortion of readout signal along the time axis, which is detected as an edge shift, may become a disturbance in performing the adjustment of the write condition with high read compatibility. For example, even when a recorded mark has any residual edge shift, if the automatic equalizer makes its tap coefficients asymmetric so that the residual edge shift can be compensated, the recorded mark will be judged such that the measured edge shift is small and hence that the recording was quite successful. In general, different drive manufacturers produce many different sorts of optical disc drives and such different optical disc drives use so many different circuit configurations. Accordingly, the recording of data in such a manner that only a particular drive can easily reproduce the recorded data will create a problem that must be solved in optical disc systems in which compatibility of recording media is highly required. The automatic equalizer of symmetric type shown in FIG. 32 can provide a solution to this problem. In FIG. 32 , a readout signal 51 reproduced from an optical disc medium (not shown) is converted to digital data by means of an A/D converter (not shown); the digital data are equalized by the automatic equalizer 22 ; and the output of the automatic equalizer 22 is then binarized by a PRML decoder 23 so that a binary bit array 52 is outputted. The tap coefficients C 0 , C 1 , C 2 . . . are subjected to an automatic learning process so that the RMS error between the target signal corresponding to the binary bit array 52 and the signal outputted from the automatic equalizer 22 can be minimized. This algorithm is usually called “LMS (Least Mean Square) Method” and performed by a LMS circuit 62 . The updated tap coefficients a 0 , a 1 , a 2 , . . . created by the LMS circuit 62 , are temporarily stored in a buffer 64 . In a work register 65 used for the actual operation of a FIR filter are set the averages each of which is the average of tap coefficients located symmetrically with each other along the time axis (e.g. combination of a 0 and a n1 , etc.). In this way, the tap coefficients of the automatic equalizer are symmetricized so that the reproduction of recorded marks with distorted edge shifts can be prevented. Incidentally, the IV converting amplifier included in a photodetector and other fulters may sometimes generate group delay due to circuit configuration. The provision, if necessary, of a group delay compensator 61 can reduce such group delay. The group delay compensator 61 can be embodied by the use of a FIR filter having asymmetric tap coefficients each of which is a preset value. Further, with this circuit configuration, it becomes possible to reduce the asymmetricity of light spot in the direction of time axis by adjusting the amount of tangential tilt in such a manner that SbER or LSEAT jitter is minimized while the data in a wellrecorded reference disc is being reproduced. With this circuit configuration, it can be made possible for the automatic equalizer to act solely on the adjustment of the frequency characteristic of reproduced signal. The automatic equalizer of symmetric type according to this invention can provide a recording condition of high reproduction compatability not only when it is combined with the LSEAT but also even when it is combined with any of conventional record adjustment methods. Since the output of the LMS circuit 62 can be transferred directly to the buffer 64 through the addition of a suitable circuit such as a selector, the automatic equalizer of symmetric type according to this invention can be easily operated as an ordinarily automatic equalizer (having no tap symmetricity restriction).
The description to follow is of the result obtained by the use of an automatic equalizer of symmetric type having 21 taps.
FIGS. 33 to 37 graphically show the experimental results regarding the adjustments of write pulse conditions by using LSEAT indices. In these experiments, the LSEAT jitters, LSEAT shifts and SbERs were measured while changing the four write pulse parameters such as Tsfp(2s, 2m), Tsfp(3s, 2m), Tsfp(2s, 3m) and Tsfp(3s, 3m) in the L0 layer of the 3layer test disc mentioned above. SbERs were measured without maintaining the symmetricity restriction on the tap coefficients of the automatic equalizer just as in the ordinary reproduction procedure. The basic unit used in adjusting the edge of write pulse was set equal to T/64, and the linear speed for writing and reading was set equal to double the speed of data transfer in case of BDs. As seen from these figures, the pulse edge positions that give the zero of LSEAT shift and the valley bottoms of LSEAT jitter and SbER, coincide with one another within a tolerance smaller than T/64. Since the adjustment unit for pulse width is usually set equal to T/16, it was ascertained from these results that the adjustment of recording condition can be performed very well by using the LSEAT shifts and the LSEAT jitters. As a result of having performed such adjustments on all the four parameters of write pulses, the value of SbER has been improved from 3×10 −3 to 1×10 −7 . FIG. 37 graphically shows the relationship obtained by measurement between write power and bit error rate. A sufficient power margin of about ±10% came to be obtained.
FIG. 38 graphically shows the relationships obtained through experiments between bit error rate and LSEAT jitter and between bit error rate and VSEAT jitter. In these experiments, the relationships between bit error rate and LSEAT jitter and between bit error rate and VSEAT jitter were measured while varying write power, defocusing, spherical aberration and tangential and radial tilts of disc media. It was ascertained from FIG. 38 that the correlation between bid error rate and jitter was further improve for LSEAT than for VSEAT. The reason for this is as described above.
The adjustment method for recording condition according to this invention will now be described on the basis of the results of the experiments and simulations described above.
FIG. 39 illustrates an example of how the adaptive parameters of write pulse are adjusted. In FIG. 39 , the adaptive parameters of write pulse are explained by the help of 4×4 type pulse tables. The results of measurement of LSEAT shifts and jitters are allocated to the 4×4 type pulse tables as described above. At this time, data are written in an optical disc medium while varying the write pulse condition, the value of the corresponding LSEAT shift is evaluated by reading out the written data, and the parameters of the write pulse are so determined as to minimize the shift value. In this way, a good condition for the write pulse can be obtained. As seen from the results shown in FIGS. 33 to 36 , adjustment results more stable against various changes come to be obtained if adjustment is performed to reach not only the condition for the minimized LSEAT shift but also the condition for the minimized LSEAT jitter. As apparent from this example, since the write pulse parameters exhibit a onetoone correspondence with the evaluation values therefor, the simultaneous adjustment of plural write pulse parameters can be parallelly performed if write/read operation is carried out while changing the plural write pulse parameters at the same time. By doing so, it becomes possible to shorten the time for test writing in a drive device to a great extent. To be concrete, although a doublespeed drive device using the method for determining the write pulse parameters one by one, takes process time of about 30 seconds to one minute, the parallel process using this method will be able to complete such test writing in about one second. In applying this adjusting method, if there are any fixed parameter among the entire write pulse parameters, adjustment can be stabilized. In general, it is preferable to fix such parameters associated with the formation of long marks as, for example, Tsfp(5s, 5m) and Telp(5s, 5m).
FIG. 40 is the flow chart illustrating the entire procedure of adjusting write pulse condition. To begin with, in Step 101 , the group delay in the reproduction circuit shown in FIG. 32 is checked, if necessary, to determine the condition for compensation of the group delay. Then, in Step 102 , the operating mode of the automatic equalizer is set to the symmetric mode. In Step 103 , while reference data are being read out, defocusing amount, spherical aberration and the tilt of disc medium are adjusted in such a manner that such indices for readout evaluation as SbER and LSEAT jitter are optimized. As described above, the tangential tilt must be adjusted with special consideration, such as by reading out plural reference data or by including the condition for optimizing the write sensitivity. In Step 104 , while the symmetrisuty, S/N ratio and crosstalk of the readout signal are being taken into consideration, proper conditions for the basic pulse and power are determined by using written data including marks and spaces having a length of 5T or greater. By doing so, the write pulse parameters for long marks corresponding to Tsfp(5s, 5m) and Telp(5s, 5m) shown in the 4×4 type pulse table is fixed. Tsfp(5s, 5m) corresponds to the write pulse condition for the anterior edge, and Telp(5s, 5m) to the write pulse condition for the posterior edge. In Steps 105 and 106 , while adaptive write pulse parameters are being adjusted, adjustment is continued until the residual edges shaft becomes less than a preset value (e.g. ±0.1% of T). In Step 107 , the performance of write pulse is evaluated by evaluating the valley bottom values for SbER and bit error rate and the power margin with respect to the obtained write pulse, and decision is made on whether a predetermined performance has been achieved. If the result of the decision indicates that the achieved performance is insufficient, the flow returns to Step 104 , where like adjustment is performed while the base pulse and the power level are being changed. When the predetermined performance has been achieved as a result of this series of steps, adjustment is finished.
FIG. 41 graphically shows the relationship obtained through experiment between focus offset and SbER. In this experiment, the automatic equalizer of symmetric type according to this invention was used. The focus offset can be adjusted to a proper value by using this relationship and making SbER minimum. The same procedure can also be used for the adjustment of redial tilt, tangential tilt, spherical aberration, etc. The Step 103 in FIG. 39 can be carried out according to this procedure.
Specific descriptions will now be provided on the adjustment method using the last pulse stating position dTLP employed by the present invention.
As described above, in the measurement results of the edge shift and the jitter based on LSEAT, the front edge of the mark is classified into a table of 4×4 constituted by four recorded mark lengths and four preceding space lengths, and the end edge of the mark is classified into a table of 4×4 constituted by four recorded mark lengths and four succeeding space lengths. The measurement results using the dTLP parameter are also classified in the same manner as the case of using LSEAT for the end edge, but recording pulses of a 2T mark include no last pulse, therefore, the measurement results using the dTLP parameter are actually classified into a table of 3×4 constituted by three recorded mark lengths (3T, 4T and 5T or greater) and four succeeding space lengths (2T, 3T, 4T and 5T or greater).
Individual values of the dTLP classified into 3×4 combination patterns are changed and written on an optical disc medium, and the written data are read out so as to evaluate a shift value of the LSEAT in every corresponding combination pattern, and each dTLP value is determined such that each shift value becomes minimum, thereby obtaining a preferable recording pulse condition. To minimize the shift value denotes to minimize an absolute value of the shift value. Specifically, the above described recording pulse adjustment method determines individual dTLP values such that each shift value of the LSEAT becomes as close to zero as possible.
FIG. 48 is a flow chart illustrating a flow of the entire dTLP adjustment. As the precondition at the start time of the adjustment, the group delay correction condition of the reproduction circuit is determined by using the above described method, and the operation mode of the automatic equalizer is set to the symmetric mode. The dTLP is set at the initial value for the recorder.
After the process is started, in Step 201 , while the reference data or the like are being read out, the defocus amount, the spherical aberration correction amount, the tilt amount of the disc medium and the like are adjusted so as to optimize the readout evaluation index.
In Step S 202 , the random data are written on the optical disc medium, and the written data are read out so as to acquire the readout signal waveforms thereof
In Step S 203 , the shift values of the LSEAT for the end edges of the mark based on the acquired readout signal waveforms are classified and calculated into 3×4 combination patterns in the above described manner.
Then, in Step S 204 , it is determined whether each of the calculated shift values of the LSEAT is minimum or not, and if the result is “No”, the dTLP value in the combination pattern that is not minimum is updated in Step S 205 . In this way, the process is repetitively executed until every shift value of the LSEAT in every combination pattern of 3×4 becomes minimum, and then the process is completed.
The above described adjustment flow can also be similarly applied to recording pulse parameters other than the dTLP.
Now, description is made of an optical disc device as an embodiment of this invention.
FIG. 1 shows in block diagram the configuration of a circuit for evaluating readout signals, the circuit being designed for realizing an optical disc device as an embodiment of this invention. In FIG. 1 , the readout signal 51 , which has been read out of an optical disc medium and passed through an analog filter not shown in the figure, is converted to a digital signal having 6 to 8 bits by an A/D converter 21 , equalized by an automatic equalizer 22 , and binarized by a PRML decoder 23 to output a binary bit array 52 . A circuit 30 for evaluating the quality of the readout signal that calculates LSEAT index, comprises main bit array detector 31 , left & right shift bit array generator 32 , ED (Euclidean Distance) difference calculator 33 , write control parameter table sorter 34 , and summary data table 35 . The main bit array detector 31 stores data corresponding to the preselected main bit arrays and judges whether a preselected main bit array is included in the binary signal 52 . When the binary signal 52 includes a preselected main bit array, the left & right shift bit array generator 32 performs the XOR process as described with, for example, FIG. 18 and generates evaluation main bit arrays for L and Rtarget signals. The ED (Euclidean Distance) difference calculator 33 calculates the Euclidean distances among the evaluation main bit arrays for the T, L and Rtarget signal and the equalized write signals 53 outputted from the automatic equalizer 22 . The write control parameter table sorter 34 statistically processes the calculated Euclidean distances in a way according to the adaptive write pulse table, in accordance with the method (expressions (D1) through (D6)) for evaluating the extended edge shifts at respective time instants or the method (expressions (7) through (13)) for evaluating the average of distributions of independently calculated Euclidean distance differences. The summary data table 35 obtains such tables as shown in, for example, FIG. 39 . The CPU 140 refers to the summary data table 35 , changes the parameters of the write pulse through controlling a write pulse adjustment circuit not shown in the figure, and adjusts the parameters of the write pulse according to the method shown in FIG. 40 .
FIG. 42 schematically shows the structure of an optical disc device incorporating the method for evaluating the readout signals according to this invention. An optical disc medium 100 mounted on the device is rotated by means of a spindle motor 160 . At the time of reading, a laserpower/pulse controller 120 controls the current flowing through the semiconductor laser 112 via the laser driver 116 in an optical head 110 so as to generate laser light 114 whose intensity is adjusted to the level instructed by the CPU 140 . The laser light 114 is focused by an objective lens 111 , to form a light spot 101 on the optical disc medium 100 . The light beam reflected from the light spot 101 is passed through the objective lens 111 to be focussed on and detected by a photodetector 113 . The photodetector 113 comprises a plurality of split photodetecting elements. A readout signal preprocessor 130 reproduces the information recorded in the optical disc medium 100 on the basis of the signal detected by the optical head 110 . The readout signal preprocessor 130 incorporates therein this invention shown as a circuit block in FIG. 1 . With this configuration, the optical disc device according to this invention can work as a device for realizing a BD system having recording density of 30 GB per disc, optimize the write pulse condition through test writing, and secure a good system margin and read compatibility.
INDUSTRIAL APPLICABILITY
Present invention relates to a method for adjusting recording condition in recording information in an highcapacity optical disc medium and an optical disc device.
DESCRIPTION OF SYMBOLS
21
A/D converter
22
automatic equalizer
23
PRML decoder
30
circuit for evaluating readout signal
31
main bit array detector
32
left & right shift bit array generator
33
ED (Euclidean Distance) difference calculator
34
write control parameter table sorter
35
summary data table
51
readout signal
52
binary bit array
53
equalized write signals
100
optical disc medium
101
light spot
110
optical head
111
objective lens
112
semiconductor laser
113
photodetector
114
laser light
115
reflected light
116
laser driver
120
laserpower/pulse controller
130
readout signal preprocessor
140
CPU
160
spindle motor