CN1694169A - Method of evaluating a readout signal, and optical disc apparatus - Google Patents

Abstract

A method of evaluating the quality of a read signal from the viewpoint of the detection margin of a Viterbi decoder in the PRML method in which a target signal level varies depending on the read signal, and an optical disc apparatus implementing the method. A method of evaluating the quality of a signal pattern comprising a combination of minimum run lengths from the viewpoint of edge shift, and an optical disc apparatus implementing the method. From the target signal level that varies depending on the read signal, a target signal is generated based on a decoding result, and an error target signal is generated in which the decoding result is edge-shifted. The signal quality is evaluated by calculating a Euclidean distance between these signals and the read signal. A virtual state that is not included in the Viterbi decoder and that is less than the minimum run length is defined, and a target signal level for the virtual state is generated using a target signal level table inside the Viterbi decoder, based on the concept of convolution. In this way, the signal quality can be evaluated by the same method as mentioned above even in cases where the pattern of a combination of the minimum run lengths has edge-shifted.

Description

The evaluation method of regenerated signal and optical disc apparatus

Technical field

The present invention relates to a kind of physical property record mark different, optical disc apparatus of recorded information of on recording medium, forming with other parts.

Background technology

Be accompanied by the high speed and the densification of CD, PRML (Partial Response MaximumLikehood) regenerated signal processing mode becomes necessary technology.Be to make the echo signal level carry out the self-adaptation PRML or the compensation PRML mode of adaptive change as one of PRML mode according to regenerated signal.In " Tech.Digest ISOM ' 03; pp.34 ", represented to use such PRML mode, by being interfered, the heat in the asymmetric of regenerated signal and when record compensates, thereby can in the device of Blu-ray correspondence, can realize being equivalent to the densification of 35GB.

In the optical disc apparatus that uses the PRML device, the same as before, suitably learn (1) regeneration equilibrium condition, (2) focal position and tilt condition, (3) recording power and impulsive condition etc. are very important.In this case, need to use the evaluation index of regenerated signal quality, various parameters are appropriately changed, be top condition.In the past, under the situation of using direct amplitude limit (グ イ レ Network ト ス ラ イ ス) regeneration, used shake to be used as the evaluation index of regenerated signal quality.As evaluation index, among " Tech.Digest ODS ' 03, pp.93 " and " the Tech.Digest ISOM ' 03; pp.116 ", corresponding to PR (1,2 corresponding to the regenerated signal quality of PRML mode, 2,1) ML channel has been represented MLSE (Maximum Likehood Sequence Error).Use is as the correct bit-array of decoded result, and with the error bit array of its displacement 1 bit, by calculating regenerated signal and both Euclidean distances (ユ one Network リ Star ド Ju From) respectively, come to estimate regenerated signal from the viewpoint of error probability.In addition, MLSE is conceived to the data edge, corresponding to the Write strategy that uses according to blank length with the suitable recording impulse condition of the long formization of mark of front and back, to each key element of form, by measuring and estimate the side-play amount of relative MLSE value, can carry out the optimization of Write strategy.In addition, in " Tech.Digest ISOM ' 03, pp.164 ",, represented PRSNR (Partial Response Signal to Noise Ratio) corresponding to PR (1,2,2,2,1) ML channel.This is at PR (1,2,2,2,1) in the ML channel, extract 3 sign indicating number shapes (パ one Application) that Euclidean distance is little and the frequency of makeing mistakes is high out, by calculating Euclidean distance respectively,, calculate regenerated signal SNR from the viewpoint of error probability from value, signal quality is estimated, represented good relevant between PRSNR and the bit error rate.

At this,, the evaluation of PRML mode error-detecting is illustrated again in order to deepen understanding to invention.

Fig. 2 is in RLL (1,7) sign indicating number, to using the sign indicating number shape of the error code under the situation that PR (1,2,2,1) class decodes, summarizes its a part of chart.In the case, because the bit number (below be called class bit number N) of performance class is 4, therefore, for the influence that 1 bit is made mistakes is considered, preferably the sign indicating number shape to 7 bits (2N-1) is considered.At this moment, will be called error code shape, extract the condition that correct sign indicating number shape and error code shape are satisfied run length (ラ Application レ Application グ ス) restriction respectively out, as shown in the figure, have 8 kinds corresponding to the combination of yard shape of 1 bit mistake with the sign indicating number shape of the central bit bit of correct sign indicating number shape.Will be to 2 powers of the difference of the echo signal level of correct sign indicating number shape and error code shape, each the time engrave the resulting and value of addition, be called both Euclidean distances, under the situation of 1 bit mistake, be (1 ²+ 2 ²+ 2 ²+ 2 ²+ 1 ²=14) 14.Be that Euclidean distance is 1.11 under 2 the situation in amplitude normalization with echo signal.About Euclidean distance, with being directed to the desired value mobile variation constantly of 2 bit code shapes, catch as the vector of M power (M=4 in this case), also this vector can be considered as 2 distance on the space of position vector.Summarizing too for 2 bit mistakes, is 12, and Euclidean distance is 14.Equally, complicated more error code shape is considered that each Euclidean distance is respectively 16,18,20,22 until infinitely.On statistics, all make a mistake in all these yards shape.But, when such signal quality that includes all error code shapes is estimated, need very big processing, be not in the CD drive can carry.Here, because Euclidean distance is divided into the distance of correct sign indicating number shape and error code shape, therefore, it is also conceivable that the index of the wrong degree of difficulty that takes place of expression.In fact, can carry out the scope of error correction, be about 10 at bit error rate for example ^-4Or 10 ^-4In the following scope, the mistake in the sign indicating number shape of minimum Euclideam distance is overriding.Therefore, if only to minimum Euclideam distance the sign indicating number shape estimate, we can say, just be enough to signal quality is estimated.MLSE just only is conceived to minimum Euclideam distance sign indicating number shape, PR is the 1T edge shift in (1,2,2,1), and the distribution that every place is produced wrong easy degree is measured, and is similar to normal distribution, and its standard deviation is estimated.

Equally, Fig. 3 and Fig. 4 represent with RLL (1,7) coding corresponding respectively, and to PR (1,2,1), error code shape and the Euclidean distance of PR (1,2,2,2,1) are summarized.

Fig. 3 is the figure that expression is summarized error code shape and the Euclidean distance of PR (1,2,1).As shown in the figure, the sign indicating number shape of minimum Euclideam distance similarly is the 1T edge shift, and Euclidean distance is 6.This is the same can the evaluation by MLSE with PR (1,2,2,1) also.

Fig. 4 represents it is the figure that error code shape and the Euclidean distance of PR (1,2,2,2,1) are summarized.As shown in the figure, the Euclidean distance of 1 bit mistake is 14, and is relative therewith, and in the sign indicating number shape of 2 bit mistakes and 3 bit mistakes, Euclidean distance is 12.In this case, only the MLSE that the 1T placement error is estimated can not correctly carry out the evaluation of signal quality.Therefore, in " Tech.Digest ISOM ' 03 pp.164 ' ", the difficulty of makeing mistakes of 3 above-mentioned sign indicating number shapes is carried out quantification, use wherein S/N minimum, the easiest sign indicating number shape of makeing mistakes, carry out the evaluation of signal quality by the S/N viewpoint.Here it is previous said PRSNR.

Fig. 5 represents it is in RLL (1,7) coding, the sign indicating number shape of the bit mistake under the situation that general employed PR (1,1,1,1) class is decoded, the figure that summarizes its part.In this case, therefore identical because class bit number N is 4 with the situation of PR (1,2,2,1) class, for the influence of 1 bit mistake is considered, can the sign indicating number shape of 7 bits be considered.As shown in the figure, to 1 bit mistake the sign indicating number shape be combined as 8 kinds, Euclidean distance is 4.For the sign indicating number shape of 10 bits, the situation that produces 2 bit mistakes is 18 kinds, and Euclidean distance is 4.Equally, when more complicated error code shape was considered, Euclidean distance was respectively 6,10 ... until infinitely.In this case, the sign indicating number shape of minimum Euclideam distance not only will be considered 1 bit mistake, also must consider 2 bit mistakes.

[non-patent literature 1] Tech.Digest ISOM ' 03, pp.34

[non-patent literature 2] Tech.Digest ODS ' 03, pp.34

[non-patent literature 3] Tech.Digest ISOM ' 03 pp.116

[non-patent literature 4] Tech.Digest ISOM ' 03 pp.164

As mentioned above, make the echo signal level, on the raising regenerability, have very big effect according to self-adaptation PRML or compensation PRML mode that regenerated signal carries out adaptive change.In addition, any one is all the same as PR (1,2,2,1) or PR (1,2,2,2,1) among above-mentioned MLSE and the PRSNR, depend on the PR class, and the echo signal level is corresponding with fixing PRML channel.

In addition, optimization viewpoint from Write strategy, use MLSE, resolve into corresponding to the long and blank long form of mark, this method of estimating regenerated signal is very desirable, but because MLSE serve as substantially with the probability of 1 bit placement error of PRML decoding, therefore, as described in above-mentioned document, can't the numeric data code shape of the mark institute structure of the blank of minimum run length and minimum run length be estimated.In existing direct amplitude limit mode is that example describes, and the repeating data sign indicating number shape of minimum run length is S/N minimum and the easiest sign indicating number shape of makeing mistakes.In addition, from the viewpoint of record controls, owing to numeric data code shape, be to interfere maximum sign indicating number shape from the heat of adjacent marker, therefore, the recording impulse condition also must be controlled it well.As mentioned above, from the viewpoint of recording/reproducing because the numeric data code shape of the combination institute structure of minimum run length is very important numeric data code shape, therefore, wish to have a kind of can be to its signal evaluation index of estimating.

Summary of the invention

Problem to be solved by this invention has following 2 points.

(problem 1) provides evaluation index and the optical disc apparatus that uses this index a kind of and the corresponding regenerated signal of PRML channel that makes the horizontal adaptive change of echo signal according to regenerated signal.

(problem 2) provides a kind of evaluation index of the regenerated signal corresponding with the data splitting sign indicating number shape of minimum run length and uses the optical disc apparatus of this index.

[problem solution]

At first, the method that solves (problem 1) is described.

MLSE can't be with the corresponding reason of target variable level, with carry out according to the target fixing horizontal as the correct bit-array of decoded result and with the calculating of the Euclidean distance of the generation of the corresponding echo signal of the error bit array of its displacement 1 bit and regenerated signal.Therefore, be benchmark with the target level that carries out adaptive change according to regenerated signal, if can carry out the generation of echo signal and the calculating of Euclidean distance, just can solve problem.About the former, from the lookup error bit-array, take out the bit-array of structural object level, by from the echo signal water-glass, loading and the corresponding echo signal level of above-mentioned bit-array one by one, just can realize.If can generate the echo signal corresponding to the target variable level, the calculating of the latter's Euclidean distance just can be by constantly obtaining at the Calais regenerated signal with 2 powers of the difference of echo signal at each mutually so.When the definition of deferring to MLSE, the difference of these 2 Euclidean distances must be carried out normalization with the benchmark Euclidean distance.At this, as the benchmark Euclidean distance, can either calculate by calculating mean value corresponding to the target level that changes, can come corresponding by the Euclidean distance that calculates the echo signal that is generated by the lookup error bit-array again.For the consideration that reduces operand, the former is better.

At this, the direction that the edge is shifted describes.Since Euclidean distance always on the occasion of, therefore can not have symbol.But, turning to the direction of the edge shift of error code shape from the deformation of correct sign indicating number, can determine each yard shape is unique.

Fig. 6 is the conclusion chart of direction of edge shift of the minimum Euclideam distance sign indicating number shape of PR (1,2,2,1).For example, be changed to from " 0001110 " under the situation of " 0000110 ", the front of judging the 3T mark is along right shift.Otherwise, be that direction of displacement is a left-hand under the situation of the error code shape " 0000110 " among the figure at playback of data.Equally, can be to the direction of all sign indicating number shape definition edge shift.Here, for example,, can expand the definition of MLSE Euclidean distance being defined as "-" under the situation of left-hand by being that Euclidean distance under the situation of dextrad is defined as "+" with direction of displacement, thereby can improve, make the directional information of edge displacement is estimated.Here, by convention, bit " 1 " expressive notation.Structure according to decoding under the situation of " 0 " expressive notation, can oppositely define direction of displacement.

Below, new signal evaluation index described herein is called S-SEAT (Signed-SequencedError for Adaptive Target).Shown in S-SEAT is defined as follows.

Be defined as ED by Euclidean distance with 2 bit code shapes " pat1 " and " pat2 " _B(pat1 pat2), is represented by following formula.

[several 1]

${\mathrm{ED}}_B(\mathrm{<figure-callout\; id="1"\; label="pat"\; filenames="A200510005117.X00731.png,A200510005117.X00801.png"\; state="\{\{state\}\}">pat</figure-callout>}1,\mathrm{pat}2)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left(V_{t\arg \mathrm{et}}\right[\mathrm{pat}1\left[n\right]]-V_{t\arg \mathrm{et}}[\mathrm{pat}2\left[n\right]\left]\right)}^2$ (formula D-1)

Here, Vtarget[B] be echo signal level for bit-array B, pat[n] bit-array among the moment n of bit code shape " pat ", N representation class bit number.

Be used for evaluation index is carried out normalized benchmark Euclidean distance, the mean value with at the Euclidean distance of 1 bit error code shape carries out to give a definition.

[several 2]

d _min＝Average(ED _min)

$=\frac{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{\mathrm{ED}}_B\left({\mathrm{Pat}}_T\right[m],{\mathrm{Pat}}_F[m\left]\right)}{M}$ (formula D-2)

Here, M is the combination sum of 1 bit error code shape, Pat _TAnd Pat _FRepresent correct bit code shape and error bit sign indicating number shape respectively.

Euclidean distance ED (pat) between regenerated signal and the designated bit sign indicating number shape " pat " is shown below.

[several 3]

$\mathrm{ED}\left(\mathrm{pat}\right)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left(V_{\mathrm{signal}}\right[t+n]-V_{t\arg \mathrm{et}}[\mathrm{pat}\left[n\right])}^2$ (formula D-3)

Wherein, Vsignal[t] represent the regenerated signal level among the moment t, 2 value results among the t are bit code shape " pat " constantly.

The value of S-SEAT is obtained by following formula as edge shift value D and the standard deviation thereof corresponding with PRML.

[several 4]

D=Sign (Shift-Direction) * { (ED (Pat _F[m])-ED (Pat _T[m]))-d _Min(formula D-4)

Sign(Shift-Direction)＝-1(Right-Edge-Shift)

(formula D-5)

+1(Left-Edge-Shift)

$\mathrm{\&sigma;}=\frac{\sqrt{\frac{\underset{p=1}{\overset{P}{\mathrm{\&Sigma;}}}{\left(D\right[p\left]\right)}^2}{\mathrm{<figure-callout\; id="2d"\; label="P"\; filenames="A200510005117.X00721.png"\; state="\{\{state\}\}">P</figure-callout>}}}}{{2d}_{\min }}$

Here, Sign (Shift-Direction) expression makes 2 values Pat as a result _TBe 1 bit mistake (edge shift), become Pat _FSituation under the edge shift direction, P represents the number of the bit code shape in computing interval of appointment.Symbol definition in (formula D-5) for negative, for just, because different with the definition of nature, so may allow the people feel strange to the edge shift of left-hand to the edge shift of dextrad, is illustrated here.In (formula D-4) (ED (PatF[m])-ED (PatT[m])), expression deducts the difference value of (Euclidean distance of regenerated signal and correct bit-array sign indicating number shape) resulting Euclidean distance from (Euclidean distance of regenerated signal and error bit array code shape).Because this value makes the decoding of correct bit code shape, therefore normally on the occasion of.Under the on all four situation of echo signal of regenerated signal and correct bit code shape, the difference value of Euclidean distance becomes benchmark Euclidean distance d _MinIn addition, be that regenerated signal is decoded to correct bit code shape and error bit sign indicating number shape with 1/2 probability under 0 the situation in this value.In addition, in the existing direct margining amplitude technique, the side-play amount between the position, edge of regenerated signal and the position, edge of clock signal is called the edge shift amount.The size of the edge shift amount in direct margining amplitude technique is that regenerated signal has 1/2 probability by mistake 2 values under 1/2 the situation of detection width (cycle of clock signal).Contrast between the two by such is shown below, and is deducting benchmark Euclidean distance d from the difference value that imports Euclidean distance _MinResulting value D ₀The time, with the edge shift amount equivalence in the direct margining amplitude technique, can handle as the edge shift amount in the PRML method.

[several 5]

D _O=(ED (Pat _F[m])-ED (Pat _T[m])-d _Min(Fill foot formula-1)

Below all be the thinking methods that follows above-mentioned MLSE value basically.As mentioned above, if regenerated signal is near the echo signal of error bit sign indicating number shape, D ₀Value is with regard to the direction of displacement that do not rely on the edge and become negative value.Therefore, must handle, make the edge shift direction consistent with the physical displacement direction of mark on being formed on CD.The edge shift direction can be according to lookup error bit code shape relatively come uniquely to determine.Therefore, among the present invention, determine the direction of edge shift,,, give D according to direction of displacement as (formula D-5) according to lookup error bit code shape ₀On on duty+1 or-1, as edge shift value D.According to (formula D-5), be under the situation on right side in the edge shift direction, the D value is under the situation in left side for just in the edge shift direction, the D value is for negative.By like this, the physical displacement direction that can make the mark that is formed on the CD is consistent with the symbol corresponding to the edge shift value D of PRML method.Here, with the edge shift direction be D value defined under the situation on right side for just, but also can carry out reciprocal definition.In this case, can the reverse symbol of (formula D-5).Directly in the margining amplitude technique, the RMS value of the edge shift amount in each edge is called jitter value, uses as the representative index of signal quality.Too, the S-SEAT value is the RMS value corresponding to the edge shift amount D of PRML method among the present invention.Be equivalent to the jitter value in the PRML method.

In addition, evaluation index of the present invention can be expanded according to situation.Below concrete extended method is described.

The echo signal level is carried out in the PRML channel of adaptive change according to regenerated signal, in order to prevent the increase of circuit scale, the calculating of the Euclidean distance of regenerated signal and echo signal, be not as substandard 2 powers of each regenerated signal constantly and echo signal with, but as absolute value and to calculate this point be effectively, detailed content will be explained below.Below such PRML channel is called absolute value series.The objective of the invention is to, the evaluation index of the regenerated signal of a kind of PRML of suiting is provided, therefore, in the PRML channel of absolute value system, can with the calculating of Euclidean distance as the absolute value of the difference of level and.Therefore, replace (formula D-1) and (formula D-3), respectively by following (formula D-7) and (formula D-8), with Euclidean distance as each in constantly both difference absolute value with calculate.

[several 6]

${\mathrm{ED}}_B(\mathrm{<figure-callout\; id="1"\; label="pat"\; filenames="A200510005117.X00731.png,A200510005117.X00801.png"\; state="\{\{state\}\}">pat</figure-callout>}1,\mathrm{pat}2)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}\left|V_{t\arg \mathrm{et}}\right[\mathrm{pat}1\left[n\right]]-V_{t\arg \mathrm{et}}[\mathrm{pat}2\left[n\right]\left]\right|$ (formula D-7)

$\mathrm{ED}\left(\mathrm{pat}\right)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}\left|V_{\mathrm{signal}}\right[t+n]-V_{t\arg \mathrm{et}}[\mathrm{pat}\left[n\right]\left]\right|$ (formula D-8)

In addition, can also replace (formula D-2), use the instantaneous value of the Euclidean distance in following (formula D-9) represented this yard shape, as benchmark Euclidean distance d _Min

[several 7]

d _Min=ED _B(Pat _T[m], Pat _F[m]) (formula D-9)

Here, ED _B(formula D-7) following by (formula D-1) calculates.

Under the situation of using (formula D-7) and (formula D-8), the echo signal level is carried out in the PRML channel of adaptive change according to regenerated signal, depend in Euclidean distance under the situation of yard shape, and since caused signal amplitude such as the sensitivity instability of medium or asymmetry value along with the time change etc. under the situation, at this moment, have and to access the advantage of evaluation of estimate quite accurately.But under the situation of circuitization such as LSI, the module of high speed motion increases, and increases this shortcoming thereby produce consumed power, therefore must be noted.Be under the situation of the PRML channel fixed, in the echo signal level because benchmark Euclidean distance d _MinAlso be fixed value, therefore, do not need to calculate d by (formula D-2) or (formula D-7) _MinValue, the constant that can use calculated in advance to come out.

Next, the calculating of the evaluation of estimate at 1 edge is investigated.About the computing method of the MLSE under the situation of RLL (1,7) coding being used PR (1,2,2,1) ML class, above-mentioned document 2 (Tech.Digest ODS ' 03, pp.34) with document 3 (Tech.Digest ISOM ' 03, on the books in pp.116).Fig. 1 and table 2 according to document 2, to the mark of the length more than 5T (T is a detection width) or the 5T and the edge of the combination institute structure of blank, also can be the same with other yards shape, according to the combination of two the sign indicating number shapes of correcting errors (being P2B/P2A and P7B/P7A in the document), calculate the evaluation of estimate (representing by MD in the document) of aforesaid edge shift.The module of concrete computing system and sign indicating number shape are on the books in the Fig. 1 of document 2 and table 1.The bit-array of 7 bits that Path A in the table 1 of document 2 and Path B are represented, from a left side in turn with document 3 P1A/P1B, P2A/P2B is corresponding.Here, be conceived to P2A (bit-array " 1110000 ") and P2B (bit-array " 1111000 ").This represents that the length of follow-up blank is more than 4T or the 4T and the back edge of the mark of long 4T of being of mark or the sign indicating number shape more than the 4T.In addition, in RLL (1, the 7) coding, the longest mark and blank length are 8T.Here, be that 8T and mark lengths are the back edge of the sign indicating number shape of 8T for the length of follow-up blank, investigate according to the block scheme of Fig. 1 of document 2.

The synoptic diagram that Figure 59 moves for the judgement situation of the sign indicating number shape detecting device of moment t and the calculating of evaluation of estimate for explanation.As shown in the figure, in t=4 and these two moment of t=5, because sign indicating number shape detecting device detects P2B and P2A respectively, so the calculating of MD value has produced 2 times.This situation can easily be judged according to the table 2 of Figure 59 and document 3, in the mark that occurs in 5T or the above length of 5T and the blank combination.If make its vague generalization more, when establishing PRML class bit number (restriction long), be created in (N+1) T or (N+1) in the combination of mark and the blank of the above length of T for N.It is factitious that physically 1 mark edge is calculated shift amount for twice.For fear of this situation, the MD value when calculating t=4 is also preserved in advance, just can not calculate when t=5.In the PRML mode of the echo signal horizontal fixed that picture PR (1,2,2,1) ML is such because the calculated value of the calculated value during t=4 during with t=5 is identical, therefore 2 calculating of evaluation of estimate are restricted to 1 time just enough.In fact, in RLL (1,7) coding, the generation frequency of the combinational code shape of mark that 5T or 5T are above and blank is less, and therefore, the influence that 2 calculating is brought is very little, in practicality, uses the MLSE value to be difficult to the generation problem in signal evaluation.

Next, to the echo signal level is investigated according to the self-adaptation PRML mode of regenerated signal adaptive change.Described in document 2, include the direction of edge shift in the definition of MLSE.In addition, shown in (formula D-5), can detect the direction of edge shift D among the present invention.Under the situation that calculates edge shift D, the direction of the edge shift during to t=4 is that the situation of left-hand is estimated among Figure 59, and the direction of the edge shift during to t=5 is that the situation of dextrad is estimated.If value separately is respectively DL and DR, in general DL and DR are values about equally.In addition, calculating by (formula D-8) under the situation such as benchmark Euclidean distance,, therefore becoming DL ≠ DR because the echo signal level changes constantly.In this case, calculate for fear of 2 times, only as above-mentioned method, limit, carry out during t=4 and calculate, do not carry out during t=5 that to calculate be not enough.But, in practicality, can think that the size difference of absolute value of DL and DR is less, for example, in RLL (1,7), if class bit number N is more than 4 or 4, the generation frequency that calculates for 2 times is just very little.As prerequisite, simplify circuit structure, under the situation of the interchangeability of the definition between attention and the MLSE, in the calculating of S-SEAT value, can be according to (formula D-6), with DL and DR addition respectively.

In addition, use like that at DVD or CD under the situation of RLL (2,10) coding, and under the class bit number is situation 3 or 3 below, the influence of calculating for 2 times becomes greatly.In addition, using minimum brigade commanders such as RLL (1,7) coding or RLL (2,10) coding is under the situation of the above coding of 3T or 3T, and as described below, in the V-SEAT that has imported virtual state (detailed content aftermentioned), 2 calculating all produces all edges.

Next, in S-SEAT and V-SEAT, avoid calculating for 2 times, the method that physically 1 edge is only calculated a shift amount describes.As follows, this method has 3 kinds.

(1) replace (formula D-4), to about the evaluation of estimate DL and situation that DR can both calculate of edge shift under, use following (formula D-10) represented average shift amount, as the shift amount D at 1 edge being had in mind.

[several 8]

$D=\frac{(\mathrm{DL}+\mathrm{DR})}{2}$ (formula D-10)

It more than is the most natural definition.

(2) replace (formula D-4), to about the evaluation of estimate DL and situation that DR can both calculate of edge shift under, use the less side's of following (formula D-11) represented such absolute value shift amount, as the shift amount D at 1 edge being had in mind.

[several 9]

D＝DL if |DL|≤|DR|

(formula D-11)

D＝DR if |DL|＞|DR|

Usually can produce error in the measurement, therefore, for example in drive assembly, under the more situation of the spike noises such as omission of clock, the method for selecting the less shift amount of absolute value like this is effective.

(3) replace (formula D-4), to about the evaluation of estimate DL and situation that DR can both calculate of edge shift under, use the bigger side's of following (formula D-12) represented such absolute value shift amount, as the shift amount D at 1 edge being had in mind.

[several 10]

D＝DR if |DL|≤|DR|

(formula D-12)

D＝DL if |DL|＞|DR|

Detailed content will be explained below, and in the PRML channel of absolute value system, it is 0 or negative value that the D0 value shown in (supplying formula-1) is merely able to.This is the feature of absolute value Series P RML channel.Because this feature, sometimes the side among DL or the DR has become 0.For fear of this situation, the method for selecting the bigger value of absolute value is effective.Above-mentioned definition is included in either party among DL or the DR and under 0 the situation, selects the opposing party.

Next, the method that solves (problem 2) is described.

Fig. 7 be corresponding to the conclusion chart of the echo signal level of PR (1,2,2,1) the ML decoding of RLL (1,7) coding.The bit number that shows the PR class under this situation is 4, and therefore, the combination of the bit-array of planting according to 24 (=16) comes the objective definition signal level.At this moment, by run-length restriction, remove the bit-array that includes the long bit of 1T, the number of significant bit array becomes 10.In the Veterbi decoding circuit, only effective bit-array is equipped with arithmetical unit.Under the situation of carrying out Viterbi decoder, the bit-array of 4 bits is divided into the bit-array of 3 bits and the bit-array of 1 bit is decoded.Though be identical in itself, just make explanation of the present invention become complicated if be described, therefore, below use the bit-array of 4 bits to describe.Underline at this, the bit-array of removing in the Veterbi decoding circuit, and do not rely on physically mark lengths, but be removed in order thoroughly to satisfy run-length restriction.

Among MLSE or the S-SEAT, the reason of evaluation that can not carry out the data splitting sign indicating number shape of minimum run length is that this is to carry out according to the structure of Viterbi decoder, as mentioned above, does not have corresponding effective bit-array in the demoder.For addressing the above problem, can only under the situation of estimating signal quality, not adopt run-length restriction, make the bit-array that includes the long bit of 1T effective, generate echo signal.

Whether effective or invalid Fig. 8 be the conclusion chart corresponding to the pairing signal level of each bit-array of Viterbi decoder portion and regenerated signal quality evaluation portion and situation.In demoder inside because run-length restriction, the significant bit array become 10 or above situation same as described above.In addition, under the situation of estimating the regenerated signal quality, by making the bit-array of getting rid of owing to run-length restriction effective, make the significant bit array get back to original 16, for example at RLL (1,7) in the coding, the front that bit-array " 0110 " (2T mark) might take place generates the echo signal that flase drop is measured the error bit array under the situation of bit-array " 0010 " (1T mark) along displacement.Under the situation of the evaluation of as above-mentioned, carrying out regenerated signal, if remove run-length restriction, determine bit-array and echo signal level, just can the combinational code shape of minimum run length be estimated.

Below, new signal evaluation index discussed above is called V-SEAT (Virtual-state-based-Sequenced Error for Adaptive Target).The computing formula of V-SEAT, identical with S-SEAT, calculate by (formula D-1) to (formula D-6) basically, can expand according to (formula D-7) to (formula D-12).V-SEAT is the same with S-SEAT, is corresponding with the variable Viterbi decoder of target level, certainly also can be corresponding with the fixing Viterbi decoder of target level.

Fig. 9 is to PR (1,2,2,1), calculates the relevant conclusion chart of the error code shape under the situation of V-SEAT.As shown in the figure,, can simplify and calculate sign indicating number shape, make it have only 4 by importing the 1T mark." X " among figure expression not only can be " 0 " but also can be " 1 ".Like this, the advantage that has the structure of brevity code shape testing circuit.

As V-SEAT, remove run-length restriction, the advantage of estimating signal quality also is present in other modes.As previously mentioned, under the situation of PR (1,2,2,2,1), the sign indicating number shape of minimum Euclideam distance owing to there is not edge shift, therefore can't be estimated signal quality by MLSE or S-SEAT.This problem all satisfies run-length restriction respectively and produces owing to praising yard shape and error code shape.For example, the sign indicating number shape No.1 to the 2 bit mistakes of Fig. 4 describes.This is that a yard shape " 0000110000000 " is made mistakes and become the situation of " 0000011000000 ", expression 2T mark right shift 1T.In fact the mark that is write down on the CD is considered.The front of supposing to have only the 2T mark is along right shift and situation about going on record.In order to revise the best titime condition, can paired pulses or power appropriately change, revise the position on the edge, front of 2T mark.In addition, the sign indicating number shape of the bit-array that Viterbi decoder is decoded, because run-length restriction, must be as described above with 2T mark former state right shift 1T.In fact just the frequency of makeing mistakes that has become with Viterbi decoder is that benchmark is estimated signal quality.But, because from resultant like this evaluation index, can only obtain the false information that the 2T mark has moved to right, therefore, if come the amendment record condition according to this evaluation index, will become the back edge that comprises entopic 2T mark, recording impulse or power condition are revised.Optical disc apparatus for implementing record regenerating can not make a mistake sometimes yet.But, for example, if by be equipped with PR (1,2,1) or directly the optical disc apparatus of amplitude limit mode regenerate, conceivablely make mistakes to being easy to.Because CD is the interchangeable storage system of medium, therefore, have to consider that the relevant issues that make such regeneration exchange no longer take place.Among the V-SEAT, signal quality is estimated, be conceived to mark edge one by one owing to surmount run-length restriction, therefore, even in these cases, also can correctly estimate the edge, front of 2T mark, it seems that from the viewpoint of recording/reproducing exchange and record controls advantage is arranged very much.Equally, for PR (1,2,2,2,1) such minimum Euclideam distance is not the PRML decoding of edge shift sign indicating number shape, also can use the evaluation index of unified edge shift benchmark, under the situation of change PRML mode, the structure of signal evaluation circuit part almost former state is fixedly used, and this advantage is very big.

Use V-SEAT,, must describe the method for solving of the echo signal level of the bit-array that includes the 1T mark for corresponding with the variable Viterbi decoder of echo signal level.In the Viterbi decoder, owing to the state that does not exist corresponding to the long bit-array of 1T, therefore, directly desired value can't be with reference to the target level table in the Viterbi decoder.The PR class, for example (1,2,2,1) are similar to the impulse response of 1T signal.The echo signal level of the PRML of fixed target is that the stack by impulse response and bit-array defines.Therefore, even for the PRML of target variable level,, just can obtain echo signal level corresponding to the bit-array that includes virtual 1T if supposition is set up by the stack of linear, additive.

Figure 10 is the bit-array of PRML of target variable level of class bit number 4 and the conclusion chart of echo signal level.(c d) represents the PRML of such target variable level for a, b by following PR.As shown in the figure, for example, the horizontal v2 of echo signal of bit-array " 0010 ", use bit-array " 0000 ", " 0001 ", the horizontal v0 of echo signal of " 0011 ", v1, v3, then consider to utilize " 0010 "=" 0011 "-" 0001 " this point, making the echo signal level of " 0000 " is not 0, can pass through

V2=v3-v1+v0 (formula 1)

Obtain.Equally, the level of other the bit-array that includes 1T can be passed through

V4=v6-v2+v0 (formula 2)

V5=v7-v2+v0 (formula 3)

V10=v8-v13+v15 (formula 4)

V11=v9-v13+v15 (formula 5)

V13=v12-v14+v15 (formula 6) obtains.

The computing formula of the echo signal level to the bit-array that includes 1T discussed above, just an example.For (formula 1), for example, can also utilize " 0010 "=" 1111 "-" 1101 " this point, by

V2=v15-v13+v0 (formula 7) obtains.Under the situation that the linear, additive between the echo signal level is set up, (formula 1) is identical with the value of (formula 7), but in general, owing to there is the effect of nonlinear heat interference etc. in the recording process, therefore, linear, additive is false.Basically by (formula 1) to (formula 6), to the echo signal level of the bit-array that includes 1T.Better method is to the value of using a plurality of computing formula to obtain, to make it that event number stack and equalization take place.Experimental result shown in the following embodiment is the mode that calculates target level by the latter.

Corresponding with the class arbitrarily that the class bit number is bigger, can adopt following method.

(method 1) calculates the target level of the isolated pulse of " 1 ".The isolated pulse of " 1 " is meant and only includes 1 " 1 " in the bit-array, in addition all is the target level of the bit-array of " 0 ".The value of these bit-array can be by 2 ⁿRepresent.Here, 0≤n＜N, N are the class bit number.The target level of these bit-array is by corresponding to the addition of the target level of run-length restriction or subtract each other and obtain.

(method 2) calculates the target level of the isolated pulse of " 0 ".The isolated pulse of " 0 " is meant and only includes 1 " 0 " in the bit-array, in addition all is the target level of the bit-array of " 1 ".The value of these bit-array can be by 2 ^N-2 ⁿRepresent.Here, 0≤n＜N, N are the class bit number.The target level of these bit-array is by corresponding to the addition of the target level of run-length restriction or subtract each other and obtain.

(method 3) obtains the target level of any bit-array.With the stack of any bit-array B, calculate by following formula as the isolated pulse of " 1 ".

[several 11]

Here, V1[B] be echo signal level for bit-array B, V _ZeroBe the target level corresponding to bit-array " 00...00 ", I1[N] for bit-array by 2 ⁿThe level of the isolated pulse of represented " 1 ", NotZero (x) is not for or not to return 1 at 0 o'clock in the value of x, is to return 0 function, “ ﹠amp at 0 o'clock; ﹠amp; " be the operational symbol of the logic product of expression integer.

Equally, with of the stack of any bit-array, calculate by following formula as the isolated pulse of " 0 ".

[several 12]

Here, V0[B] be echo signal level for bit-array B, V _OneBe the target level corresponding to bit-array " 11...11 ", I0[N] for bit-array by 2 ^N-2 ⁿThe level of the isolated pulse of represented " 0 ", IsZero (x) is for being to return 1 at 0 o'clock in the value of x, is not to return 0 function at 0 o'clock.

Will be according to (formula 8), (formula 9), the target level equalization of obtaining as the stack of the pulse of the pulse of " 1 " and " 0 " is obtained target level by following formula respectively.

[several 13]

$V\left[B\right]=\frac{\left(V1\right[B]+V0[B\left]\right)}{2}$ (formula 10)

The same with S-SEAT, V-SEAT also can adapt to above-mentioned expansion.Concrete as above use (D-7) as described in the explanation of (D-12).

Figure 11 is the definition of MLSE, S-SEAT, V-SEAT and the conclusion chart of feature.

More than solution of (problem 1) of the present invention and (problem 2) is illustrated.Among the following embodiment, to being elaborated with the different corresponding concrete method and the experimental result of PR class.

Below, among the present invention, if do not express especially, the calculating of S-SEAT value and V-SEAT value as the RMS value of all values of the edge shift D that calculates equally with the MLSE value, is concluded explanation to expanded definition.In addition, same, if do not express especially, the PRML channel is 2 powers systems.

The optical disc apparatus that uses regenerated signal evaluation method provided by the present invention and use this method, can carry out: the regenerated signal evaluation that (1) is corresponding with the PRML channel that makes the horizontal adaptive change of echo signal according to regenerated signal, the evaluation of the corresponding regenerated signal of data splitting sign indicating number shape of (2) and minimum run length.

Description of drawings

Fig. 1 is the synoptic diagram of the structure example of the S-SEAT counting circuit installed in the explanation optical disc apparatus of the present invention.

Fig. 2 is in RLL (1,7) coding, and the part of the bit error code shape under the situation that use PR (1,2,2,1) class is decoded is concluded chart.

Fig. 3 is in RLL (1,7) coding, the error code shape of PR (1,2,1) class and the conclusion chart of Euclidean distance.

Fig. 4 is in RLL (1,7) coding, the error code shape of PR (1,2,2,2,1) class and the conclusion chart of Euclidean distance.

Fig. 5 is in RLL (1,7) coding, the error code shape of PR (1,1,1,1) class and the conclusion chart of Euclidean distance.

Fig. 6 is in RLL (1,7) coding, the conclusion chart of the direction of the edge shift of the minimum Euclideam distance sign indicating number shape of PR (1,2,2,1) class.

Fig. 7 be corresponding to the conclusion chart of the echo signal level of PR (1,2,2, the 1) class of RLL (1,7) coding.

Fig. 8 be corresponding to the conclusion chart of the pairing signal level of each bit-array of PR (1,2,2,1) the Veterbi decoding portion of RLL (1,7) coding and regenerated signal quality evaluation portion and whether effective or invalid situation.

Fig. 9 is to PR (1,2,2,1), calculates the relevant conclusion chart of the error code shape under the situation of V-SEAT.

Figure 10 is the bit-array of PRML of target variable level of class bit number 4 and the conclusion chart of echo signal level.

Figure 11 is the definition of MLSE, S-SEAT, V-SEAT and the conclusion chart of feature.

Figure 12 is about encode for RLL (1,7) pairing PR (1,2,1) and PR (a, b, c) the conclusion chart of the direction of the Euclidean distance of 1 bit error code shape of class and edge shift.

Figure 13 is the variable PR of the pairing target level of RLL (1,7) coding (a, b, c) the conclusion chart of the echo signal level of ML decoding.

Figure 14 is for RLL (1,7) encode pairing PR (1,2,1) and PR (a, b, c) the conclusion chart of the error detecting code shape of the V-SEAT of class and edge shift direction.

Figure 15 is the conclusion chart of the echo signal level of fixing PR (1,2,1) the ML decoding of the pairing target level of RLL (1,7) coding.

Figure 16 is the variable PR of the pairing target level of RLL (1,7) coding (a, b, c) the conclusion chart of the echo signal level of ML decoding.

Figure 17 is about encode for RLL (1,7) pairing PR (1,2,2,2,1) and PR (a, b, c, d, e) the conclusion chart of the direction of the Euclidean distance of 1 bit error code shape of class and edge shift.

Figure 18 is the variable PR of the pairing target level of RLL (1,7) coding (a, b, c, d, e) the conclusion chart of the echo signal level of ML decoding.

Figure 19 is for RLL (1,7) encode pairing PR (1,2,2,2,1) and PR (a, b, c, d, e) the conclusion chart of the error detecting code shape of the V-SEAT of class and edge shift direction.

Figure 20 is the conclusion chart of the echo signal level of fixing PR (1,2,2,2,1) the ML decoding of the pairing target level of RLL (1,7) coding.

Figure 21 is the variable PR of the pairing target level of RLL (1,7) coding (a, b, c, d, e) the conclusion chart of the echo signal level of ML decoding.

Figure 22 is about encode for RLL (1,7) pairing PR (1,2,3,3,2,1) and PR (a, b, c, d, e, f) the conclusion chart of the direction of the Euclidean distance of 1 bit error code shape of class and edge shift.

Figure 23 is the variable PR of the pairing target level of RLL (1,7) coding (a, b, c, d, e, f) the conclusion chart of the echo signal level of ML decoding.

Figure 24 is for RLL (1,7) encode pairing PR (1,2,3,3,2,1) and PR (a, b, c, d, e, f) the conclusion chart of the error detecting code shape of the V-SEAT of class and edge shift direction.

Figure 25-1 is the conclusion chart of the echo signal level of fixing PR (1,2,3,3,2,1) the ML decoding of the pairing target level of RLL (1,7) coding.

Figure 25-2 is the conclusion chart of the echo signal level of fixing PR (1,2,3,3,2,1) the ML decoding of the pairing target level of RLL (1,7) coding.

Figure 26-1 is the variable PR of the pairing target level of RLL (1,7) coding (a, b, c, d, e, f) the conclusion chart of the echo signal level of ML decoding.

Figure 26-2 is the variable PR of the pairing target level of RLL (1,7) coding (a, b, c, d, e, f) the conclusion chart of the echo signal level of ML decoding.

Figure 27 is about encode for RLL (1,7) pairing PR (1,2,2,1) and PR (a, b, c, d) the conclusion chart of the direction of the Euclidean distance of 1 bit error code shape of class and edge shift.

Figure 28 be the variable PR of the pairing target level of RLL (1,7) coding (a, b, c, d) the conclusion chart of the echo signal level of ML decoding.

Figure 29 is for RLL (1,7) encode pairing PR (1,2,2,1) and PR (a, b, c, d) the conclusion chart of the error detecting code shape of the V-SEAT of class and edge shift direction.

Figure 30 is the conclusion chart of the echo signal level of fixing PR (1,2,2,1) the ML decoding of the pairing target level of RLL (1,7) coding.

Figure 31 is the variable PR of the pairing target level of RLL (1,7) coding (a, b, c, d) the conclusion chart of the echo signal level of ML decoding.

Figure 32 is about encode for RLL (2,10) pairing PR (3,4,4,3) and PR (a, b, c, d) the conclusion chart of the direction of the Euclidean distance of 1 bit error code shape of class and edge shift.

Figure 33 be the variable PR of the pairing target level of RLL (2,10) coding (a, b, c, d) the conclusion chart of the echo signal level of ML decoding.

Figure 34 is for RLL (2,10) encode pairing PR (3,4,4,3) and PR (a, b, c, d) the conclusion chart of the error detecting code shape of the V-SEAT of class and edge shift direction.

Figure 35 is the conclusion chart of the echo signal level of fixing PR (3,4,4,3) the ML decoding of the pairing target level of RLL (2,10) coding.

Figure 36 is the variable PR of the pairing target level of RLL (2,10) coding (a, b, c, d) the conclusion chart of the echo signal level of ML decoding.

Figure 37 for the pairing target level of explanation RLL (1,7) coding will variable PR (a, b, c) the ML echo signal level of decoding be assumed to asymmetric amount 0 and situation about obtaining under the synoptic diagram of embodiment.

Figure 38 for the pairing target level of explanation RLL (1,7) coding will variable PR (a, b, c, d) the ML echo signal level of decoding be assumed to asymmetric amount 0 and situation about obtaining under the synoptic diagram of embodiment.

Figure 39 for the pairing target level of explanation RLL (2,10) coding will variable PR (a, b, c, d) the ML echo signal level of decoding be assumed to asymmetric amount 0 and situation about obtaining under the synoptic diagram of embodiment.

Figure 40 is the synoptic diagram of the structure example of the V-SEAT counting circuit installed in the explanation optical disc apparatus of the present invention.

Figure 41 is the synoptic diagram of another structure example of the S-SEAT counting circuit installed in the explanation optical disc apparatus of the present invention.

Figure 42 is the synoptic diagram of another structure example of the S-SEAT counting circuit installed in the explanation optical disc apparatus of the present invention.

Figure 43 is the synoptic diagram of another structure example of the S-SEAT counting circuit installed in the explanation optical disc apparatus of the present invention.

Figure 44 is the synoptic diagram of the structure example of the S-SEAT that installed in the explanation optical disc apparatus of the present invention and V-SEAT counting circuit.

Figure 45 is to (1+D) ⁿSeries, the test findings synoptic diagram of the relation between declare record capacity and the bit error rate.

Figure 46 be to (1,2 ..., 2,1) and series, the test findings synoptic diagram of the relation between declare record capacity and the bit error rate.

Figure 47 is the approximate series of paired pulses response, the experimental result synoptic diagram of the relation between declare record capacity and the bit error rate.

Figure 48 is the synoptic diagram of the embodiment of the key concept of the information reproduction method of explanation offset-type PRML mode.

Figure 49 is the experimental result synoptic diagram of the difference of the high capacity performance of explanation offset-type PRML mode and other PRML mode.

Figure 50 is the synoptic diagram of the structure example of explanation optical disc apparatus of the present invention.

When Figure 51 uses PR (3,4,4,3) class regeneration DVD-RAM dish for explanation, MLSE, S-SEAT, the synoptic diagram of the measured result of V-SEAT.

Figure 52 for explanation use PR (3,4,4,3) and PR (a, b, c, d) during class regeneration DVD-RAM dish, by MLSE, S-SEAT, V-SEAT is to the result's that estimates by the edge shift amount of the blank sign indicating number shape of distinguishing with the relation of mark in front and back mode chart.

The synoptic diagram of flow process is write in the examination that Figure 53 appropriately changes the condition of recording impulse for explanation.

The example that Figure 54 writes as the examination of using S-SEAT of the present invention or V-SEAT for explanation, the synoptic diagram of the experimental result that the displacement amount and the relation between each evaluation of estimate of recording impulse are measured.

Figure 55 is an example of writing as the examination of using S-SEAT of the present invention or V-SEAT, and the synoptic diagram of the experimental result of the relation between balanced recruitment and bit error rate and each evaluation of estimate is described.

Figure 56 for explanation use PR (1,2,2,1) and PR (a, b, c, d) during class regeneration trial-production dish, MLSE, S-SEAT, the synoptic diagram of the measured result of V-SEAT.

Figure 57 for explanation use PR (1,2,2,1) and PR (a, b, c, d) during class regeneration trial-production dish, by MLSE, S-SEAT, V-SEAT is to the result's that estimates by the edge shift amount of the blank sign indicating number shape of distinguishing with the relation of mark in front and back synoptic diagram.

Figure 58 is the example as the academic probation of using V-SEAT of the present invention, and the synoptic diagram of the experimental result of the relation between balanced recruitment and bit error rate and the V-SEAT value is described.

The synoptic diagram that Figure 59 moves for the judgement situation of the sign indicating number shape detecting device of moment t and the calculating of evaluation of estimate for explanation.

When Figure 60 uses PR (3,4,4,3) class regeneration DVD-RAM dish for explanation, MLSE, S-SEAT, the synoptic diagram of the measured result of V-SEAT.

Figure 61 for explanation use PR (a, b, c, d) during class regeneration DVD-RAM dish, MLSE, S-SEAT, the synoptic diagram of the measured result of V-SEAT.

The PR that Figure 62 for explanation use absolute value is (a, b, c, when d) class is regenerated the DVD-RAM dish, MLSE, S-SEAT, the synoptic diagram of the measured result of V-SEAT.

Figure 63 for explanation use PR (a, b, c, d) during class regeneration DVD-RAM dish, the synoptic diagram of the different caused V-SEAT measured result that edge portion value is handled.

The PR that Figure 64 for explanation use absolute value is (a, b, c, when d) class is regenerated the DVD-RAM dish, the synoptic diagram of the different caused V-SEAT measured result of edge portion value processing.

Among the figure: the 10... decoding unit, 11... waveform equalizer, 12... branch metric calculation unit, 13...ACS unit, 14... path memory, 15...PR object table, 16... sign indicating number shape compensation meter, 17... target level table, the conditional target level table of 18..., 19... pattern control module, 191... switch, 20... target level unit, 21... target level computing unit, 22... Error Calculation and equalization unit, 24... sign indicating number shape detecting device, 25... equalization unit, 26... Error Calculation and action control unit, 27... switch, 30... the signal evaluation unit, 31... sign indicating number shape selected cell, 32... target level computing unit, 33... target level computing unit, 34... the sequence error evaluation unit, 35... virtual target level calculation unit, 50... regenerated signal, 51...2 value result, 52... calibration value, the correct sign indicating number of 53... shape, 54... error code shape, 55... evaluation of estimate, 100... CD, 101... luminous point, 110... shaven head, 111... object lens, 112 semiconductor lasers, 113... photodetector, 120... record data controller, 130... regenerated signal processor, 140...CPU, 150... servo controller, 160... spindle drive motor, 170... interface, the 180... main frame

Embodiment

The present invention is described in detail below by embodiment.

[embodiment 1]

Corresponding with various PR classes

In the above explanation, to RLL (1,7) corresponding PR (1,2,2, the 1) decoding of encoding, set forth computing method about S-SEAT and V-SEAT.Below, expression is corresponding to the PR (1,2,1) of RLL (1,7) coding, PR (12221), and PR (123321) class, and corresponding to the embodiment of PR (3,4,4, the 3) class of RLL (2,10).

At first, to setting forth with the corresponding PR class of RLL (1, the 7) coding of blu-ray disc etc.

Figure 12 is to PR (1,2,1) and PR (a, b, c) 1 bit error code shape of class, the figure that summarize the direction of Euclidean distance and edge shift corresponding with RLL (1,7) coding.As shown in the figure, for 1 bit mistake the sign indicating number shape be combined as 2 kinds, under the situation of fixed target level, Euclidean distance is 6.For the definition of the edge shift direction of each yard shape according to diagram.

At first, the calculating to S-SEAT describes.

Figure 13 is the variable PR of the pairing target level of RLL (1,7) coding (a, b, c) the conclusion chart of the echo signal level of ML decoding.Under the situation of calculating S-SEAT, can extract the bit code shape of Figure 12 out, calculate evaluation of estimate according to above-mentioned definition.

Secondly, the calculating to V-SEAT describes.

Figure 14 is for PR (1,2,1) and PR (a, b, c) the conclusion chart of the error detecting code shape of the V-SEAT of class and edge shift direction corresponding with RLL (1,7) coding.As shown in the figure, for 1 bit mistake the sign indicating number shape be combined as 4 kinds, under the situation that is the fixed target level, Euclidean distance is 6.To the definition of the edge shift direction of each yard shape according to diagram.

Figure 15 is the conclusion chart of the echo signal level of fixing PR (1,2,1) the ML decoding of the pairing target level of RLL (1,7) coding.Here, the echo signal level is normalized to ± 1 describe.Under the situation of calculating V-SEAT, can extract the bit code shape of Figure 14 out, calculate evaluation of estimate according to above-mentioned definition.

Figure 16 is the variable PR of the pairing target level of RLL (1,7) coding (a, b, c) the conclusion chart of the echo signal level of ML decoding.Under the situation of calculating V-SEAT, can calculate corresponding to the echo signal level that includes the long bit-array of 1T according to the definition among the figure, extract the bit code shape of Figure 14 out, calculate evaluation of estimate according to above-mentioned definition.

Figure 17 is about encode for RLL (1,7) pairing PR (1,2,2,2,1) and PR (a, b, c, d, e) the conclusion chart of the direction of the Euclidean distance of 1 bit error code shape of class and edge shift.As shown in the figure, for 1 bit mistake the sign indicating number shape be combined as 18 kinds, under the situation that is the fixed target level, Euclidean distance is 14.To the definition of the edge shift direction of each yard shape according to diagram.

At first, the calculating to S-SEAT describes.

Figure 18 is the variable PR of the pairing target level of RLL (1,7) coding (a, b, c, d, e) the conclusion chart of the echo signal level of ML decoding.Under the situation of calculating S-SEAT, can extract the bit code shape of Figure 17 out, calculate evaluation of estimate according to above-mentioned definition.

Secondly, the calculating to V-SEAT describes.

Figure 19 is for RLL (1,7) encode pairing PR (1,2,2,2,1) and PR (a, b, c, d, e) the conclusion chart of the error detecting code shape of the V-SEAT of class and edge shift direction.As shown in the figure, for 1 bit mistake the sign indicating number shape be combined as 4 kinds, under the situation that is the fixed target level, Euclidean distance is 14.To the definition of the edge shift direction of each yard shape according to diagram.

Figure 20 is the conclusion chart of the echo signal level of fixing PR (1,2,2,2,1) the ML decoding of the pairing target level of RLL (1,7) coding.Here, the echo signal level is normalized to ± 1 describe.Under the situation of calculating V-SEAT, can extract the bit code shape of Figure 19 out, calculate evaluation of estimate according to above-mentioned definition.

Figure 21 is the variable PR of the pairing target level of RLL (1,7) coding (a, b, c, d, e) the conclusion chart of the echo signal level of ML decoding.Under the situation of calculating V-SEAT, can calculate corresponding to the echo signal level that includes the long bit-array of 1T according to the definition among the figure, extract the bit code shape of Figure 19 out, calculate evaluation of estimate according to above-mentioned definition.

Figure 22 is about encode for RLL (1,7) pairing PR (1,2,3,3,2,1) and PR (a, b, c, d, e, f) the conclusion chart of the direction of the Euclidean distance of 1 bit error code shape of class and edge shift.As shown in the figure, for 1 bit mistake the sign indicating number shape be combined as 18 kinds, under the situation that is the fixed target level, Euclidean distance is 28.To the definition of the edge shift direction of each yard shape according to diagram.

At first, the calculating to S-SEAT describes.

Figure 23 is the variable PR of the pairing target level of RLL (1,7) coding (a, b, c, d, e, f) the conclusion chart of the echo signal level of ML decoding.Under the situation of calculating S-SEAT, can extract the bit code shape of Figure 22 out, calculate evaluation of estimate according to above-mentioned definition.

Next, the calculating to V-SEAT describes.

Figure 24 is for RLL (1,7) encode pairing PR (1,2,3,3,2,1) and PR (a, b, c, d, e, f) the conclusion chart of the error detecting code shape of the V-SEAT of class and edge shift direction.As shown in the figure, for 1 bit mistake the sign indicating number shape be combined as 4 kinds, under the situation that is the fixed target level, Euclidean distance is 28.To the definition of the edge shift direction of each yard shape according to diagram.

Figure 25 is the conclusion chart of the echo signal level of fixing PR (1,2,3,3,2,1) the ML decoding of the pairing target level of RLL (1,7) coding.At this, the echo signal level is normalized to ± 1 describe.Under the situation of calculating V-SEAT, can extract the bit code shape of Figure 24 out, calculate evaluation of estimate according to above-mentioned definition.

Figure 26 is the variable PR of the pairing target level of RLL (1,7) coding (a, b, c, d, e, f) the conclusion chart of the echo signal level of ML decoding.Under the situation of calculating V-SEAT, can calculate corresponding to the echo signal level that includes the long bit-array of 1T according to the definition among the figure, extract the bit code shape of Figure 24 out, calculate evaluation of estimate according to above-mentioned definition.At this, expression is by above-mentioned formula (formula 8), and (formula 9) calculates after the target level to " 1 " pulse and " 0 " pulse, by (formula 10), calculates the situation of other target level.The target level of " 1 " pulse is v2, v4, v8, v16, and the target level of " 0 " pulse is v61, v59, v55, v47.

As PR class, again PR (1,2,2,1) is concluded explanation at last corresponding to RLL (1,7) coding.Though repeat with above-mentioned explanation, more deep for the understanding of the present invention, be described below.

Figure 27 is about encode for RLL (1,7) pairing PR (1,2,2,1) and PR (a, b, c, d) the conclusion chart of the direction of the Euclidean distance of 1 bit error code shape of class and edge shift.As shown in the figure, for 1 bit mistake the sign indicating number shape be combined as 8 kinds, under the situation that is the fixed target level, Euclidean distance is 10.To the definition of the edge shift direction of each yard shape according to diagram.

At first, the calculating to S-SEAT describes.

Figure 28 be the variable PR of the pairing target level of RLL (1,7) coding (a, b, c, d) the conclusion chart of the echo signal level of ML decoding.Under the situation of calculating S-SEAT, can extract the bit code shape of Figure 27 out, calculate evaluation of estimate according to above-mentioned definition.

Next, the calculating to V-SEAT describes.

Figure 29 is for RLL (1,7) encode pairing PR (1,2,2,1) and PR (a, b, c, d) the conclusion chart of the error detecting code shape of the V-SEAT of class and edge shift direction.As shown in the figure, for 1 bit mistake the sign indicating number shape be combined as 4 kinds, under the situation that is the fixed target level, Euclidean distance is 10.To the definition of the edge shift direction of each yard shape according to diagram.

Figure 30 is the conclusion chart of the echo signal level of fixing PR (1,2,2,1) the ML decoding of the pairing target level of RLL (1,7) coding.Here, the echo signal level is normalized to ± 1 describe.Under the situation of calculating V-SEAT, can extract the bit code shape of Figure 29 out, calculate evaluation of estimate according to above-mentioned definition.

Figure 31 is the variable PR of the pairing target level of RLL (1,7) coding (a, b, c, d) the conclusion chart of the echo signal level of ML decoding.Under the situation of calculating V-SEAT, can calculate corresponding to the echo signal level that includes the long bit-array of 1T according to the definition among the figure, extract the bit code shape of Figure 29 out, calculate evaluation of estimate according to above-mentioned definition.

Next, the pairing PR class of RLL (2,10) coding to CD/DVD describes.

Figure 32 is about encode for RLL (2,10) pairing PR (3,4,4,3) and PR (a, b, c, d) the conclusion chart of the direction of the Euclidean distance of 1 bit error code shape of class and edge shift.As shown in the figure, for 1 bit mistake the sign indicating number shape be combined as 2 kinds, under the situation that is the fixed target level, Euclidean distance is 50.To the definition of the edge shift direction of each yard shape according to diagram.

At first, the calculating to S-SEAT describes.

Figure 33 be the variable PR of the pairing target level of RLL (2,10) coding (a, b, c, d) the conclusion chart of the echo signal level of ML decoding.Under the situation of calculating S-SEAT, can extract the bit code shape of Figure 32 out, calculate evaluation of estimate according to above-mentioned definition.

Next, the calculating to V-SEAT describes.

Figure 34 is for RLL (2,10) encode pairing PR (3,4,4,3) and PR (a, b, c, d) the conclusion chart of the error detecting code shape of the V-SEAT of class and edge shift direction.As shown in the figure, for 1 bit mistake the sign indicating number shape be combined as 4 kinds, under the situation that is the fixed target level, Euclidean distance is 50.To the definition of the edge shift direction of each yard shape according to diagram.

Figure 35 is the conclusion chart of the echo signal level of fixing PR (3,4,4,3) the ML decoding of the pairing target level of RLL (2,10) coding.Here, the echo signal level is normalized to ± 1 describe.Under the situation of calculating V-SEAT, can extract the bit code shape of Figure 34 out, calculate evaluation of estimate according to above-mentioned definition.

Figure 36 is the variable PR of the pairing target level of RLL (2,10) coding (a, b, c, d) the conclusion chart of the echo signal level of ML decoding.Under the situation of calculating V-SEAT, can calculate corresponding to the echo signal level that includes the long bit-array of 1T length and 2T according to the definition among the figure, extract the bit code shape of Figure 35 out, calculate evaluation of estimate according to above-mentioned definition.

In the foregoing description about V-SEAT, be not 0 to the asymmetric of regenerated signal, " 1 " pulse is that unequal this general situation is illustrated with the size of " 0 " pulse.Among the following embodiment, asymmetric less to regenerated signal regarded " 1 " pulse under the situation about equating as with the size of " 0 " pulse that the method for solving of the target variable level of Jian Danhuaing is illustrated more.

Among the present invention, for convenience's sake, with the PR class of target variable level for example be expressed as PR (a, b, c, d).Asymmetric in supposition is under 0 the situation, moment variation that will " 1 " impulse response be normalized to (c d), is normalized to signal amplitude ± 1 for a, b, then " 0 " impulse response become (a ,-b ,-c ,-d).If according to the target level table, obtain a respectively, b, c, the value of d just can calculate the echo signal level to any bit-array by the stack amount computing of bit-array and impulse response, therefore, can simplify processing.Here, will have a that is obtained according to the target level table, b, c, the PR class of d value is called convolution PR class.Below, illustrated respectively PR corresponding to RLL (1,7) coding (a, b, c), PR (a, b, c, d), and corresponding to PR (a, b, c, the embodiment of each class d) of RLL (2,10) coding.

Figure 37 to will variable PR with the pairing target level of RLL (1,7) coding (a, b, c) the ML echo signal level of decoding be assumed to asymmetric amount 0 and situation about obtaining under be illustrated.Here, the echo signal level is normalized to ± 1.For obtaining convolution PR class, as shown in the figure, by

A={ (v1-v0)-(v6-v7) }/2 (formulas 11)

B={ (v3-v1)-(v4-v6) }/2 (formulas 12)

C={ (v4-v0)-(v3-v7) }/2 (formulas 13) obtain.With the big or small equalization of " 1 " pulse, just can easily calculate respectively with reference to Figure 16 with " 0 " pulse.Being normalized to 0 with asymmetric, is under ± 1 the situation with amplitude normalization, v0=+1, V7=-1.It is utilized, just can calculate for example such target level arbitrarily of v1=v0+a.The method for solving of all target levels is according to diagram.

Figure 38 to will variable PR with the pairing target level of RLL (1,7) coding (a, b, c, d) the ML echo signal level of decoding be assumed to asymmetric amount 0 and situation about obtaining under be illustrated.Here, the echo signal level is normalized to ± 1.For obtaining convolution PR class, as shown in the figure, by

A={ (v1-v0)-(v14-v15) }/2 (formulas 14)

B={ (v3-v1)-(v12-v14) }/2 (formulas 15)

C={ (v6-v2)-(v9-v13) }/2 (formulas 16)

D={ (v8-v0)-(v7-v15) }/2 (formulas 17) obtain.With the big or small equalization of " 1 " pulse, just can easily calculate respectively with reference to Figure 31 with " 0 " pulse.Asymmetric be 0, be under ± 1 the situation with amplitude normalization, v0=+1, V15=-1.It is utilized, just can calculate for example such target level arbitrarily of v1=v0+a.The method for solving of all target levels is according to diagram.

Figure 39 to will variable PR with the pairing target level of RLL (2,10) coding (a, b, c, d) the ML echo signal level of decoding be assumed to asymmetric amount 0 and situation about obtaining under be illustrated.Here, the echo signal level is normalized to ± 1.For obtaining convolution PR class, as shown in the figure, by

A={ (v1-v0)-(v14-v15) }/2 (formulas 18)

B={ (v3-v1)-(v12-v14) }/2 (formulas 19)

C={ (v6-v2)-(v9-v13) }/2 (formulas 20)

D={ (v8-v0)-(v7-v15) }/2 (formulas 21) obtain.These formula are all identical to (formula 17) with (formula 14).With the big or small equalization of " 1 " pulse, just can easily calculate respectively with reference to Figure 26 with " 0 " pulse.Asymmetric be 0, be under ± 1 the situation with amplitude normalization, v0=+1, V15=-1.It is utilized, just can calculate for example such target level arbitrarily of v1=v0+a.The method for solving of all target levels is according to diagram.

In addition under the situation, for PR (a, b, c, d, e), PR (a, b, c, d, e f) waits too, respectively with the big or small equalization of " 1 " pulse with " 0 " pulse, calculate convolution PR class according to the echo signal water-glass, use this convolution PR class, can be to any bit-array objective definition level.Such result, with relative asymmetric be not echo signal level under 0 the situation, be the same with " 1 " of bit-array with the big or small equalization that " 0 " exchanges resulting bit-array level.This operation corresponding to asymmetric be 0 situation.

[embodiment 2]

Secondly, about being used for calculating the appropriate circuit structure of S-SEAT and V-SEAT, the contrast accompanying drawing describes embodiment.

The structure of the S-SEAT counting circuit of being installed in the optical disc apparatus of the present invention has been described among Fig. 1.All structures are made of decoding unit 10, target level unit 20 and signal evaluation unit 30.

At first decoding unit 10 is described.Decoding unit 10 is made of waveform equalizer 11, branch metric calculation unit 12, ACS (Add Select Compare) unit 13, path memory 14 (パ ス メ モ リ) and target level table 17.Regenerated signal 50 is converted to digital value by AD converter in advance, after the enforcement of the FIR wave filter in the waveform equalizer 11 equilibrium treatment, in branch metric calculation unit 12, calculate 2 powers (branched measurement value) of the error of desired value and each bit-array.The desired value of each bit-array is with reference to the numerical value that is accommodated in the target level table 17.In the ACS unit 13, in state before 1 moment and the metric each state under (along with the migration over-and-over addition companion branched measurement value of state, and apply to handle it is not dispersed), add branched measurement value corresponding to each bit-array.At this moment, proceed among the transition process (be generally 2,, be 1 sometimes) of the state of current time, select the less processing of metric according to run-length restriction.The bit-array that state representation is preserved the migration in 1 moment, for example the PR analogy specially for 4 situation under, be 4 bits by bit-array, state is that 3 bits are represented.In the path memory 14, the 2 value results that each bit-array is decoded preserved with the sufficiently long time, along with the renewal of time, carried out the shifting processing of memory content, made it always preserve up-to-date result.ACS unit 13, when the selection of transition process is handled, according to selection result the information of being preserved in the path memory is arranged again, by carrying out such processing repeatedly, gradually the information in the path memory is integrated, thereby after the sufficiently long time, in which bit-array, all become same value, finished so-called path and merged.2 value results 51 are when constantly upgrading, the 2 value information of being taken out from path memory.

Secondly, target level unit 20 is described.Target level unit 20 is made of sign indicating number shape detecting device 24 and equalization unit 25.Sign indicating number shape detecting unit 24,2 value results 51 of storage class number of bits will send to equalization unit 25 corresponding to the address information of this bit-array.Equalization unit 25, the level of the regenerated signal that will be exported from waveform equalizer 11, to each should be constantly address information, promptly to each corresponding averaging of bit-array, go forward side by side and be about to that it is kept at the processing of the memory table of the correspondence in target level table 17.

At last, signal evaluation unit 30 is described.Signal evaluation unit 30, by sign indicating number shape selected cell 31, target level computing unit 32 and 33 and sequence error evaluation unit 34 constitute.Sign indicating number shape selected cell 31, corresponding to 1 bit mistake, it is individual that 2 value results 51 are preserved above-mentioned " class bit number * 2-1 ", judges whether the calculating sign indicating number shape into S-SEAT.Be judged as under the situation of calculating sign indicating number shape, it as correct sign indicating number shape 53, is being sent to target level computing unit 32, simultaneously, generating above-mentioned 1 bit error code shape, it as error code shape 54, is being sent to target level computing unit 33.In the target level computing unit 32 and 33, reference object water-glass 17, output is corresponding to correct sign indicating number shape 53 and corresponding to the echo signal level of error code shape 54.In the sequence error evaluation unit 34, use the output of target level computing unit 32 and 33, and the output of waveform equalizer 11, according to (formula D-1) definition, calculate S-SEAT value 55 to (formula D-6).As mentioned above, S-SEAT value 55 can also be expanded according to (formula D-7) to (formula D-12).

The structure of having represented the V-SEAT counting circuit installed in the optical disc apparatus of the present invention among Figure 40.All structures are made of decoding unit 10, target level unit 20 and signal evaluation unit 30.Elemental motion is same as the previously described embodiments.Calculate peculiar textural element as V-SEAT, added virtual target level calculation unit 35.In the virtual target level calculation unit 35, be not subjected to the restriction of run length, use each tabular value of target level table 17, carry out the processing that calculates and store to the echo signal level of all bit-array by said method.In addition, in the sign indicating number shape selected cell 31, select processing corresponding to the bit code shape of V-SEAT.In the sequence error evaluation unit 34, use the output of target level computing unit 32 and 33, and the output of waveform equalizer 11, according to (formula D-1) definition, calculate V-SEAT value 55 to (formula D-6).As mentioned above, V-SEAT value 55 can also be expanded according to (formula D-7) to (formula D-12).

At last, signal evaluation unit 30 is described.Signal evaluation unit 30, by sign indicating number shape selected cell 31, target level computing unit 32 and 33 and sequence error evaluation unit 34 constitute.Sign indicating number shape selected cell 31, corresponding to 1 bit mistake, it is individual that 2 value results 51 are preserved above-mentioned " class bit number * 2-1 ", judges whether the calculating sign indicating number shape into S-SEAT.Be judged as under the situation of calculating sign indicating number shape, it as correct sign indicating number shape 53, is being sent to target level computing unit 32, simultaneously, generating above-mentioned 1 bit error code shape, it as error code shape 54, is being sent to target level computing unit 33.In the target level computing unit 32 and 33, reference object water-glass 17, output is corresponding to correct sign indicating number shape 53 and corresponding to the echo signal level of error code shape 54.In the sequence error evaluation unit 34, use the output of target level computing unit 32 and 33, and the output of waveform equalizer 11, according to (formula D-1) definition, calculate S-SEAT value 55 to (formula D-6).As mentioned above, S-SEAT value 55 can also be expanded according to (formula D-7) to (formula D-12).

Among 2 above embodiment, to target level unit 20, under the situation of carrying out the signal regeneration processing, example to always action is illustrated, but under the situation of the quality of estimating regenerated signal, the method that target level is fixing is very desirable in the stable this point of estimating.Therefore, under the situation of estimating signal quality, preferably stop the action of target level unit 20, the value of target level table 17 is not upgraded.

Another structure of having represented the S-SEAT counting circuit installed in the optical disc apparatus of the present invention among Figure 41.With the above embodiments different target level units 20 that are structurally.In the present embodiment, target level computing unit 21, Error Calculation and action control unit 26, switch 27 have also been added in the target level unit 20.In the variable PRML decoding of target level, upgrade the value of target level table 17 according to regenerated signal.In order to ensure the stability of the action of optical disc apparatus, must limit implementing to upgrade situation about handling.For example, in (1) search, (4) though under the unsettled situation of action not shown but PLL (the Phase Locked Loop) circuit of extraction clock from regenerated signal, (3) there are dust, fingerprint, bad point etc. in the CD media, under the situation that regenerated signal significantly is out of shape, handle by the various renewals of implementing the echo signal level, so-called entanglement state then can occur, hinder the regular event of demoder.For fear of above-mentioned item, must testing circuit be set respectively to above-mentioned various situations, stop to upgrade and handle.Method shown in the present embodiment, always calculate the error of echo signal and regenerated signal,, implement the renewal of echo signal level and handle because an absolute value in error is under setting or the situation below the setting, therefore, can realize above-mentioned functions simply.Move as described below.In target level computing unit 21, import 2 value results 51, always the export target signal level.In Error Calculation and action control unit 26, output to waveform equalizer 11, estimate with the error of target level computing unit 21, only the absolute value in error is under setting or the situation below the setting, Closing Switch 27, implement the renewal of echo signal level and handle, control like this.In the present embodiment, in order to deepen understanding to invention, imported switch 27, but, for example make the clock ON/OFF that offers equalization unit 25 move control method, or in the deferent segment of equalization unit 25, add logic integrated circuit, it is easier to move the method for control by this circuit, is preferably in and adopts these methods in the actual circuit.

Below, to in the PRML of target variable level mode, improving very outstanding offset-type PRML mode on the recording density this point, and be used for improving the target level restricted type PRML mode that performance is exchanged in regeneration, the embodiment of its operating principle and S-SEAT of the present invention and V-SEAT counting circuit is described.

Improve the PRML mode, realize the problem of the high capacity significantly of CD, caused non-linear edge shift such as the heat interference when being to suppress to write down.As previously mentioned, the PRML method is determined echo signal by the selection of PR class.Here, to the different signal of record recording density in 1 CD, the difference of densification performance is discussed according to various PR classes.

The CD of being prepared is that convolution phase change film is resulting on the substrate of the back-trough structure of orbit interval 0.34 μ m.In the experiment, the DDU-1000 type disc evaluation device that uses パ Le ス テ Star Network company to be produced.The wavelength of light source is 405nm, and the NA of object lens is 0.85.Use RLL (1,7) as modulating-coding, detection window width Tw changes in the scope of 53nm to 80nm.The recording capacity of the single face of the CD of CD size under the situation of Tw=53nm, is 35GB.

After the difference of PR class discussed, select following 3 series.

(1) (1+D) ⁿSeries

Be the most basic class series, be PR (1,1), PR (1,2,1), PR (1,3,3,1) ...

(2) (1,2 ..., 2,1) series

Be the series of the PR (1,2,2,1) that includes more use in CD, compare that high boost reduces, and can expect the improvement of S/N ratio with above-mentioned series.

(3) the approximate series of impulse response

The PR class is similar to the impulse response of reproducing head substantially.

Here, use the optical analogy device to calculate the equipment of the pulse of shaven head, in the PR class, use.

Use selected each PR class, use the desirable regenerated signal that is calculated by the optical analogy device, the decision equilibrium condition and the CD signal of regenerating make the target of each PR class and the RMS error minimum between the regenerated signal.The tap number of balanced device is 11.

Figure 44 to Figure 46 has represented the measurement result to the regenerability of the CD of each PR series respectively.

Represented to be directed to (1+D) among Figure 45 (a) ⁿSeries, the relation of recording capacity and bit error rate.Figure 45 (b) is the conclusion chart of the upper limit of bit performance, significant bit number of arrays, number of significant condition, pinpoint target number of levels and the recording capacity of each PR class.The upper limit of recording capacity represents that bit error rate is 10 ^-4Or 10 ^-4Following scope.If class bit number (the key element number that is comprised in the performance of PR class) is N, then the number of all bit-array just is 2 ^N,, from the set of bit-array, remove minimum run length and be that resultant bit-array number is exactly the significant bit number of arrays after the 1T according to run-length restriction.Number of significant condition etc. are obtained too.After having realized aforesaid operations, because circuit scale is directly proportional with the significant bit number of arrays, hope has a kind ofly can make the as far as possible little mode of class bit number.In this series, the class bit number is big more, just can carry out densification more more, but the class bit number reaches 6 or 6 when above, and performance improves just saturated.Be that the dominant record capacity is 31GB under 7 the situation at the class bit number.

Represented among Figure 46 (a) for (1,2 ... 2,1) series, the relation of recording capacity and bit error rate.Figure 46 (b) concludes chart for details.In this series, the class bit number is excessive, and recording capacity will reduce.The class bit number is big more, just can show the time-shift of regenerated signal finer and smoothlyer, but simultaneously since independently the target level number also become many, therefore, consider that the difference to the target level in two different paths diminishes, mistake will increase during routing.Be that this serial dominant record capacity is 31GB under 5 the situation at the class bit number.

Represented among Figure 47 (a) for the approximate series of impulse response, the relation of recording capacity and bit error rate.Figure 47 (b) concludes chart for details.Too, if the class bit number is excessive, recording capacity will reduce in this series.Be that this serial dominant record capacity is 32GB under 5 the situation at the class bit number.

According to 3 kinds of results that PR class series is discussed to considering, knowing only increases PR class bit number, can make structure complicated, also restricted on performance improves simultaneously.This be because, in the regenerated signal of CD, exist owing to interfere between the caused coding of shape of luminous point, and the heat during record is interfered caused non-linear edge shift.For with such non-uniform encoding between interfere and edge shift corresponding, determine this basic PRML method scarce capacity of desired value by the stack computing of linearity, must be someway by other, the compensating non-linear composition.According to The above results, in order to realize further densification, below 2 very important.

(1), make the target level number not increase by not increasing the class bit number.

(2) in stack computing determined desired value, add compensation rate corresponding to bit-array, the compensation desired value, by come like this with regenerated signal in the non-linear component that comprised carry out corresponding.

Realize high capacity in order to satisfy above-mentioned condition, can adopt the determined desired value of stack computing with the NN bit, interpolation is corresponding to the N (compensation rate of the bit-array of bit of N＞NN), the decision desired value, itself and regenerated signal are compared, the most correct in the bit-array with the N bit simultaneously, i.e. the PRML mode of bit-array 2 values of the error minimum of regenerated signal and desired value.

Figure 48 is the embodiment of the key concept of the above-mentioned information reproduction method of explanation.For explanation is simplified, describe for example with the most basic class PR (1,1).Mode 1 is the most basic PRML mode.Shown in structure example, the desired value that will be corresponding with continuous 2 constantly bit-array and the value of regenerated signal compare, and select the bit-array of error minimum.In this example, the target level number is 3, can't with the asymmetric and nonlinear coding of regenerated signal between interfere carry out corresponding.

Mode 2 is to be published in Technical Digest of ISOM 2002, the self-adaptation PRML mode among the 269-271 (2002).To by in the determined desired value of stack computing, add offset V corresponding to the bit-array of 2 bits, to use as new desired value, the bit-array of the error minimum between the value of selection and regenerated signal is carried out 2 values.The number of offset V is 4 (=2 ²).Can be according to regenerated signal asymmetric, desired value is changed adaptively, interfere but can not eliminate fully between nonlinear coding.

Mode 3 owing to give the front and back of PR (1,1) bit-array, is added the sign indicating number shape compensation bit of 1 bit respectively, therefore is called offset-type PRML.Be characterised in that, in desired value, added the pairing offset V of bit-array of 4 bits of the sign indicating number shape compensation bit different with mode 2.On this basis, will compare corresponding to the desired value and the regenerated signal of the bit-array of 4 bits, the bit-array of Select Error minimum is carried out 2 values.In this mode, the number by stack computing determined target level remains 3 not to be increased, and the number of offset V is 16 (=2 ⁴), therefore, can be in the scope of the bit-array of 4 bits interfere between to non-uniform encoding to compensate.In order to distinguish mutually, the performance of PR class is described as Compensated-PR (0,1,10) or CPR (0,1,1,0) with former PRML method.This is that the class bit number is 4 PRML method, desired value is identical with former explanation, by coefficient row (0,1,1,0) row calculate with the stack computing of the bit-array of 4 bits, but because the coefficient of each 1 bit of two ends is 0, therefore, coefficient row (1,1) the determined desired value with 2 bits is identical.In addition, front and back coefficient " 0 " indication code shape compensation bit, the meaning of CPR becomes the determined offset V of bit-array that adds 4 bits in desired value.By same method, when representing mode 1 in the past, then become PR (1,1), mode 2 can be described as CPR (1,1).

Experimental result among the figure is to carry out record under the condition of above-mentioned CD detection width Tw=57nm (memory capacity is 32.5GB), by the born again result of mode separately.Here, basic PR class is PR (1,2,2,1), and the transfer rate of data is 100Mbps.Bit error rate can access 50 * 10 under the situation of mode 1 (PR (1,2,2,1)) ^-4, under the situation of mode 2 (CPR (1,2,2,1)), can access 15 * 10 ^-4, under the situation of mode 3 (CPR (0,1,2,2,1,0)), can access 0.5 * 10 ^-4According to mode 3, the acknowledgement bit rate can be below 1/100 or 1/100.In addition, the eye pattern of regenerated signal, the actual effect signal (compensational regeneration signal) under the situation of each mode of expression use can learn that eyes have opened brightly in the mode 3.The S/N ratio of the signal of the 2Tw that is comprised in the compensating signal is 3.6dB in mode 1, is 6.1dB in mode 2, is 9.5dB in the manner.

Figure 49 (a) is for illustrating in offset-type PRML mode and other PRML modes experimental result picture of the difference of high capacity performance.Select PR (1,2,2,1) as basic PR class.If the allowable value of bit error rate is 10 ^-4, can obtain the upper limit of recording capacity.Being limited among the PR (1,2,2,1) on the recording capacity of mode in the past is 30GB, and CPR is 32GB in (1,2,2,1).Being limited among CPR (0,1,2,2,1, the 0) ML4 on the recording capacity of offset-type PRML mode is 32.5GB, and CPR (0,1,2,2,1,0) in 34.5GB, CPR (0,0,1,2,2,1,0,0) and CPR (0,0,0,1,2,2,1,0,0,0) in 35GB or more than the 35GB.CPR (0,1,2,2,1,0) ML4 represents, only determines offset by 6 bits, the mode that bit number (ML bit number) maintenance 4 bits of selecting bit-array the most accurately to carry out best decoding are carried out.Though compared advantage with former technology, do not carry out best decoding processing owing to do not comprise a yard shape compensation bit, therefore, the inhibition ability of non-linear displacement reduces.In order to bring into play the optimum capacity of offset-type PRML mode to greatest extent, it is very important including yard best decoding processing of shape compensation bit.Here resulting result, basic PR class is not limited in PR (1,2,2,1), and comparing with above-mentioned various PR classes further to increase recording capacity.

Figure 49 (b) is the conclusion chart to the bit-array number of each mode shown in Figure 49 (a), status number, number of levels, sign indicating number shape compensation bit number and ML bit number.Owing to be used for realizing that the circuit scale of PRML mode is directly proportional with the bit-array number substantially, therefore, increase the CPR (0 of 3 bits respectively in order to realize yard shape compensation bit front and back, 0,0,1,2,2,1,0,0,0), need with PR (1,2,2,1) is in a ratio of circuit scale more than 10 times or 10 times, obtains the balanced extremely important of performance and circuit scale.

Here, select PR (1,2,2,1) as basic PR class, the mode that front and back is increased the sign indicating number shape compensation bit of similar number describes.But the present invention is not limited to this.As basic PR class, can select PR (1,1), PR (1,2,1), PR (3,4,4,3), PR (1,1,1,1), PR (1,2,2,2,1) etc., any basic PR class.The situation of symmetry can also look like CPR (0,1,2,2,1), CPR (0,0,1,2,2,1), CPR (1,2,2,1,0), CPR (1,2,2,1,0,0) like this before and after sign indicating number shape compensation bit number was not limited in addition, added asymmetrical bit number.For example, the influence that the heat when sometimes writing down is interfered concentrates on the front side edge edge, if physically bear physically signal very clearly again, so only adding sign indicating number shape compensation bit in the front side is optimal selection.

Figure 42 is the another kind of structure of the S-SEAT counting circuit installed in the explanation optical disc apparatus of the present invention.This structure is corresponding with offset-type PRML.

Decoding unit 10 is made of waveform equalizer 11, branch metric calculation unit 12, ACS unit 13, path memory 14, PR object table 15 and sign indicating number shape compensation meter 16.Regenerated signal 50 is converted to digital value by AD converter in advance, after the enforcement of the FIR wave filter in the waveform equalizer 11 equilibrium treatment, in branch metric calculation unit 12, calculate 2 powers (branched measurement value) of the error of desired value and each bit-array.As this moment desired value, with the corresponding initial target value of bit-array with reference to PR object table 15, corresponding to the offset reference code shape compensation meter 16 of bit-array, use both resulting results of value addition.The action of ACS unit 13 and path memory 14 as hereinbefore.The key of present embodiment is, adds offset as new desired value to desired value, is result the most accurately by coming 2 values like this.

Target level unit 20 is made of target level computing unit 21 and Error Calculation and equalization unit 22.In the target level computing unit 21, with 2 value results as input, with the corresponding initial target value of bit-array with reference to PR object table 15, with the corresponding offset reference code of bit-array shape compensation meter 16, with both value addition as the echo signal horizontal output.Error Calculation and equalization unit 22 calculate the error of the output of the output of waveform equalizer 11 and target level computing unit 21, to each bit-array equalization margin of error, implement to upgrade the processing of the value of sign indicating number shape compensation meter 16.

Calculating about S-SEAT, can make target level computing unit 31 and 32 carry out work, will with the corresponding initial target value of bit-array with reference to PR object table 15, with the corresponding offset reference code of bit-array shape compensation meter 16, as the echo signal level, the calculating of S-SEAT value as hereinbefore with both additions.

Include the variable PRM mode of target level of offset-type PRML mode, can decode, improve regenerability corresponding to regenerated signal.In addition, for example cause under the situation of luminous point distortion, and exist in the I-V amplifier of regenerated signal under the situation of distortion such as group delay, regenerate corresponding to these distortion at tangential tilt.Therefore, when in the optical disc apparatus of the PRML mode of the fixed target level that other are installed, regenerating, must consider the deterioration of regenerated signal quality.As mentioned above, the problem points of the variable PRM mode of target level is to guarantee the interchangeability of regenerating.In order to address this problem, not the target level that changes all bit-array corresponding to regenerated signal, but, must allow not servo-actuated of target level impairing the key element of regeneration interchangeability.One of implementation is, measures the asymmetric amount of regenerated signal, uses according to the predetermined target level table of asymmetric amount.Two of implementation is that making the target level of the combination of the symmetrical bit-array of relative time counter-rotating is identical value.As mentioned above, because any one of the representation example of regenerated signal distortion all is to make regenerated signal asymmetrical deformation on time orientation, therefore, for example the echo signal level of the echo signal level of bit-array " 1000 " and bit-array " 0001 " dissimilates.Therefore, by will with the pairing echo signal horizontal of the contrast bit-array homogenizing of relative these time reversals, become same value, the echo signal level is not followed the tracks of the distortion of the regenerated signal of time orientation, thereby can improve regeneration interchangeability.The PRML mode of implementing these restrictions is called target level restricted type PRML mode.

Figure 43 is the another kind of structure of the S-SEAT counting circuit installed in the explanation optical disc apparatus of the present invention.This structure is corresponding with target level restricted type PRML.

Decoding unit 10 is made of target level table 18, pattern control module 19 and the switch 191 of waveform equalizer 11, branch metric calculation unit 12, ACS unit 13, path memory 14, target level table 17, band restriction.Regenerated signal 50 is transformed to digital value by AD converter in advance, after the enforcement of the FIR wave filter in the waveform equalizer 11 equilibrium treatment, in branch metric calculation unit 12, each bit rate is calculated branched measurement value.The either party of the value in the value in the use target level table 17 or the target level table 18 of band restriction is as the desired value of this moment.Target level table 17 changes corresponding to regenerated signal as previously mentioned, and the value of the target level table 18 of band restriction as previously mentioned, makes it become same value the averaging of target level of the contrast bit array group of relative time counter-rotating.The action of 19 pairs of switches 191 of pattern control module is controlled, and which target level control uses.For example when learn focus is offset, when the examination of carrying out the learning records condition is write and when confirming the checking of regenerated signal quality, use the target level table 18 of band restriction.Under the situation of in data reproduction, makeing mistakes, use the value of target level table 17.When common data reproduction, use which target level can.The selection action of such echo signal level of being undertaken by pattern control module 19 can improve regeneration interchangeability and improve learning accuracy.

Target level unit 20 is made of sign indicating number shape detecting device 24, equalization unit 25 and restriction control module 23.The action of sign indicating number shape detecting unit 24 and equalization unit 25, and the change of target level table 17 order is with aforesaid identical.Restriction control module 23 receives the output of equalization unit 25, and averaging processing makes the target level of bit-array group of the contrast of relative time counter-rotating become same value, and this result is kept in the target level table 18 that band limits.

Calculating about S-SEAT, can make target level computing unit 31 and 32 carry out work, by the judgement of pattern control module 19, the value of using target level table 17 or being with the target level table 18 that limits is as the target level value corresponding to bit-array, and the computing method of value as hereinbefore.

In the present embodiment, in order to help understanding, to come the example of option table to be illustrated by pattern control module 19 and switch 191 to invention.Target level table 17 and the target level table 18 of being with restriction, only therefore the value difference of the target level of preserving to some extent has identical hardware configuration.Here, in order to simplify circuit structure, can an installation targets water-glass 17, by the not control of the CPU of expression among the figure, each value of backup in advance in S-RAM or D-RAM is loaded some values in the target level table 17 according to necessity.Like this, pattern control module 19 has not just needed with switch 191, can dwindle circuit scale.

More than circuit structure of the present invention is illustrated.Figure 41 extremely embodiment shown in Figure 43 is the structure of S-SEAT counting circuit.In order to make it corresponding, can increase the virtual target level calculation unit 35 shown in Figure 40, with the option code deformation of sign indicating number shape selected cell 31 V-SEAT more with V-SEAT.

At last, the circuit structure that has the function of calculating S-SEAT and V-SEAT by a circuit is described.

The S-SEAT that installed in the optical disc apparatus of the present invention and the structure of V-SEAT counting circuit have been represented among Figure 44.With Figure 40 different signal evaluation unit 30 that are structurally.

Signal evaluation unit 30 is made of sign indicating number shape selected cell 31, target level computing unit 32,33, sequence error evaluation unit 34, virtual target level calculation unit 35, S-SEAT sign indicating number shape table 36 and V-SEAT sign indicating number shape table 37.Sign indicating number shape selected cell 31 according to the not indication of the CPU of expression among the figure, switches the computing function of S-SEAT and V-SEAT.Specifically, when calculating S-SEAT, with reference to the S-SEAT sign indicating number shape table 36 of preserving S-SEAT bit code shape, option code shape, output correct sign indicating number shape 53 and error code shape 54 are carried out work like this.When calculating V-SEAT, with reference to the V-SEAT sign indicating number shape table 37 of preserving V-SEAT bit code shape, option code shape, output correct sign indicating number shape 53 and error code shape 54 are carried out work like this.After output correct sign indicating number shape 53 and the error code shape 54, later calculating does not need to change by S-SEAT and V-SEAT just as hereinbefore.Though be unnecessary, but the influence of the virtual target level calculation unit 35 when in advance S-SEAT being calculated describes, because S-SEAT sign indicating number shape is satisfied run-length restriction, therefore, not with reference to for the elimination in the virtual target level calculation unit 35 the echo signal level of bit-array of run-length restriction, just can not be subjected to above-mentioned influence.The echo signal level that satisfies run-length restriction is identical under the situation of S-SEAT and V-SEAT.Therefore, in this structure, virtual target level calculation unit 35 can not bring bad influence when S-SEAT calculated, and can calculate correct S-SEAT and V-SEAT.

In the present embodiment, the PRML demoder corresponding to the fixed target level has been described, the circuit structure that carries out the calculating of S-SEAT and V-SEAT.In order to make the variable PRML demoder of itself and target level corresponding, can be with reference to the structure of Figure 41 to Figure 43, signal evaluation unit 30 is changed to equipment in the present embodiment.

[embodiment 3]

Implement the situation of branch metric calculation by absolute value

As previously mentioned, in the Viterbi decoder, in order to obtain 2 value results the most accurately, use 2 powers with the difference between regenerated signal and the desired value resulting branched measurement value that multiplies each other.Such Viterbi decoder is called 2 powers system.2 power Δs 2 of the difference of regenerated signal and target level become:

[several 14]

Δ ²＝(V _signal[t]-V _target[n]) ²

(formula A-1)

＝(V _signal[t]) ²-2V _signal[t]V _target[n]+(V _target[n]) ²

Here, Vsignal[t] the expression regenerated signal level of t constantly, Vtarget[n] expression is corresponding to the echo signal level of bit-array.In Viterbi decoder, carry out 2 values and make Δ ²The value that multiplies each other becomes minimum.The 1st of the right is the level of regenerated signal, owing to being common to all bit-array, does not therefore need to calculate.Under the situation of echo signal level, can carry out following computing for fixing Viterbi decoder.

[several 15]

Δ ²=A[n] V _Signal[t]+B[n] (formula A-2)

Here, A[n]=-2Vtarget[n], B[n]=(Vtarget[n]) 2 be respectively constant.At A[n] value be under 1,2,4 the situation replace multiplying each other owing to can use bit to be shifted, therefore, under a lot of situations, compare with (formula A-1), (A-2) can simplify operand significantly, also be circuit scale.

In addition, target level is in the variable Viterbi decoder, because target level is variable, therefore is merely able to be reduced to:

[several 16]

Δ ²=-2V _Signal[t] V _Target[n]+(V _Target[n]) ²(formula A-3)

Because Vtarget[n] be variable, the simplification that replaces with the circuit structure that bit is shifted of therefore will multiplying each other is very difficult.Therefore, in the variable Viterbi decoder of target level, circuit scale must not be not very big.

For it is simplified, the absolute value of the difference by using regenerated signal and target level, rather than 2 powers are used as branched measurement value, are effectively.At this moment, as

[several 17]

| Δ |=| V _Signal[t]-V _Target[n] | (formula A-4)

Use | Δ |, by carrying out Veterbi decoding, compare with (formula A-3), owing to do not need to multiply each other, therefore, the bit number of arithmetical unit can be less, thereby can realize the reduction with consumed power dwindled of circuit scale.

Below the employed S-SEAT of Veterbi decoding and the V-SEAT computing method of absolute value system are concluded.

About error detecting code shape, absolute value system can use identical method with 2 power classes.

In the calculating of S-SEAT, can be to define Euclidean distance by absolute value.Shown in being defined as follows.

Be defined as ED by Euclidean distance with 2 bit code shapes " pat1 " and " pat2 " ^B(pat1 pat2), is represented by following formula, and is identical with aforesaid (formula D-7).

[several 18]

${\mathrm{ED}}_B(\mathrm{<figure-callout\; id="1"\; label="pat"\; filenames="A200510005117.X00731.png,A200510005117.X00801.png"\; state="\{\{state\}\}">pat</figure-callout>}1,\mathrm{pat}2)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}\left|V_{t\arg \mathrm{et}}\right[\mathrm{pat}1\left[n\right]]-V_{t\arg \mathrm{et}}[\mathrm{pat}2\left[n\right]\left]\right|$ (formula D-13)

Here, V _Target[B] expression is for the echo signal level of bit-array B, pat[n] bit-array among the moment n of expression bit code shape " pat ", N representation class bit constant.

With average minimum Euclideam distance, be defined as the mean value of the Euclidean distance of 1 bit error code shape or below the mean value, this is identical with (formula D-2).

[several 19]

d _min＝Average(ED _min)

$=\frac{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{\mathrm{ED}}_B\left({\mathrm{Pat}}_T\right[m],\mathrm{Pa}{t}_F[m\left]\right)}{M}$ (formula D-14)

Here, M is the sum of 1 bit error code shape combination, Pat _TAnd Pat _FRepresent correct bit code shape and error bit sign indicating number shape respectively.

Euclidean distance ED (pat) between the bit code shape " pat " of regenerated signal and appointment represents by following formula.

[several 20]

$\mathrm{ED}\left(\mathrm{pat}\right)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}\left|V_{\mathrm{signal}}\right[t+n]-V_{t\arg \mathrm{et}}[\mathrm{pat}\left[n\right]\left]\right|$ (formula D15)

Here, V _Signal[t] expression is the regenerated signal level of t constantly, and 2 value results among the t are bit code shape " pat " constantly.

The value σ of S-SEAT obtains by following formula.This is identical to (formula D-6) with (formula D-4).

[several 20]

D=Sign (Shift-Direction) * { (ED (Pat _F[m])-ED (Pat _T[m]))-d _Min(formula D-16)

Sign(Shift-Direction)＝-1(Right-Edge-Shift)

(formula D-17)

+1(Left-Edge-Shift)

$\mathrm{\&sigma;}=\frac{\sqrt{\frac{\underset{p=1}{\overset{P}{\mathrm{\&Sigma;}}}{\left(D\right[p\left]\right)}^2}{\mathrm{<figure-callout\; id="2d"\; label="P"\; filenames="A200510005117.X00721.png"\; state="\{\{state\}\}">P</figure-callout>}}}}{{2d}_{\min }}$ (formula D-18)

Here, Sign (Shift-Direction) expression makes 2 values PatT1 bit mistake (edge shift) as a result, becomes the edge shift direction under the situation of PatF, and P represents the number of the bit code shape in computing interval of appointment.

Because the definition of V-SEAT is identical with S-SEAT, therefore, can use (formula D-13) to (D-18), calculates V-SEAT.About error detecting code shape, identical with shown in the 2 powers systems.

As previously mentioned, S-SEAT value and V-SEAT value can be expanded according to formula (D-8) to (D-12).

Is that the example of Viterbi decoder describes by the following examples to absolute value

[embodiment 4]

Experimental result

At first, as suitable example, the experimental result of using the DVD-RAM dish is described RLL (2,10) coding.

When having represented use PR (3,4,4,3) regeneration DVD-RAM dish among Figure 51, the measured result of MLSE, S-SEAT and V-SEAT.CD is 2 times of commercially available fast DVD-RAM media.Evaluating apparatus is the LM330A that シ バ ソ Network company is produced, and optical maser wavelength is 658nm, and numerical aperture is 0.60.About regenerative loop, use the equilibrium condition (3 tap FIR wave filters and No. 6 low-pass filters) of DVD-RAM standard, as balanced device.Recruitment is 5.5dB.To recording power is 10.3mW, and removing power is 4.7mW, rewrites 10 times in the groove track, and the regenerated signal quality of each mode is estimated.Jitter value is 8.5%.Each evaluation of estimate is the Gaussian distribution shape as shown in the figure, and transverse axis is as detection width, for ± Tw/2.According to the definition of this evaluation of estimate, surpassed ± edge of the scope of Tw/2 switches and just produces the regeneration mistake, therefore, can equally with jitter value handle.Each evaluation of estimate is respectively MLSE=11.0%, S-SEAT=11.0%, V-SEAT=12.5%.The distribution of MLSE therefrom mind-set left avertence moves, and mainly is because the size of the 3T signal in the regenerated signal is littler than the echo signal level of PR (3,4,4,3) ML class.

Figure 60 is to above mensuration, makes the adaptive equalizer of the FIR wave filter institute structure of 5Tap carry out result under the situation of work.It is 13.2% that the jitter value of regenerated signal worsens, and this is to carry out the result of automatic equalization near the echo signal of PR (3,4,4,3) ML class.The measured result of MLSE, S-SEAT, V-SEAT is compared with the situation of not carrying out automatic equalization and is diminished, and shows and improves effect.

Figure 61 is to identical mensuration, uses PR (a, b, c, the result under the situation of d) regenerating.Adaptive equalizer does not carry out work.In the calculating of MLSE value, with the PR (3,4,4,3) of fixed target level as echo signal.Improved the S-SEAT value, diminishing is 7.3%.In addition, the fixedly use of V-SEAT value and Figure 51 fixedly situation of PR (3,4,4,3) ML class is compared change greatly, is because as previously mentioned, and V-SEAT carries out 2 times to each edge and calculates.

Figure 62 is to identical mensuration, uses PR (a, b, c, the result under the situation of d) regenerating of absolute value system.Adaptive equalizer does not carry out work.Because the MLSE value does not define the PRML channel of absolute value system, therefore, replaces said method here, expresses incongruent S-SEAT value.Incongruent S-SEAT value in definition (formula D-5), is calculated identically.

[several 22]

Sign(Shift-Direction)＝1

Incongruent S-SEAT value is for the definition of the fixed target class of 2 powers system, and is identical with MLSE.Relative 0 skew of the distribution of incongruent S-SEAT value is the feature of absolute value system at left demifacet, and this point as previously mentioned.Owing to depart from purport of the present invention, therefore it is not elaborated, but during according to the calculating definition of branched measurement value, can proves significantly to become such distribution expressly.Relative 0 of S-SEAT value and V-SEAT value are distributed in the right and left, according to the symbol of definition (formula D-5), under the situation of edge right shift, write picture on the right side at relative 0.According to above-mentioned example, S-SEAT of the present invention and V-SEAT also can be corresponding for the PRML channel of absolute value system.

Figure 63 is to identical mensuration, because the synoptic diagram of the variation of the different caused V-SEAT value that edge portion handles.Use PR (a, b, c, d) the ML class is regenerated, and adaptive equalizer does not carry out work.As previously mentioned, because the V-SEAT evaluation index imported virtual state,, calculate and shift left and by the evaluation of estimate of displacement therefore to all edges.As previously mentioned, the processing of the measurement result of the evaluation of estimate of edge shift has 4 kinds, just as shown in the figure.About the size of value, select the bigger all value＞selection mean value＞less values of selection of value＞selection in order, the value of RMS diminishes, and can access the result of nature.In evaluation of estimate, under the situation as for example phase place comparative result of PLL (Phase Locked Loop) circuit, preferably select mean value with edge shift of the present invention.If select common edge shift evaluation of estimate, for example under the less situation of relatively poor situation of S/N and minimum brigade commander's signal amplitude, to the command voltage of VCO (Voltage Controlled Oscilator) easily about (frequency UP/DOWN direction) vibration.In addition, if use according to the evaluation of estimate of the edge shift of V-SEAT, even in these cases, owing to be conceived to the equilibrium of left and right sidesing shifting, therefore can be stablized the command voltage to VCO, obtain good clock signal.In the optical disc apparatus described later, also can use such PLL circuit.

Figure 64 is to identical mensuration, because the synoptic diagram of the variation of the different caused V-SEAT value that edge portion handles.The PR of use absolute value system (a, b, c, d) the ML class is regenerated, and adaptive equalizer does not carry out work.Identical with the tendency of The above results, but select mean value as the minimum this point difference of RMS value under the situation of the evaluation of estimate of edge shift.This be because, as previously mentioned, in the PRML channel of absolute value system, branched measurement value is necessary for Euclidean distance or below the Euclidean distance, does not have the S-SEAT value of symbol to be distributed in 0 left side.By this feature, for the distribution that makes the S-SEAT value is the center with 0, left-right symmetric distributes, thereby to the method for the averaging of evaluation of estimate of the edge shift of left and right directions, is the result who becomes littler than each minimum value.

Figure 52 is in above mensuration, estimates the result of the edge shift amount of the sign indicating number shape of distinguishing by the blank and the mark of front and back by MLSE, S-SEAT, V-SEAT.Among the figure, and SFP (s, m) the expression front is along displacement, and s is the length of leading blanks, and m is the length of this mark.Equally, (s is the length of follow-up blank to ELP for s, m) expression back edge displacement, and m is the length of this mark.This definition to the edge displacement is described below.As previously mentioned, for MLSE that is subjected to run-length restriction and S-SEAT, the front can't be estimated along displacement SFP (3,3) and back edge displacement ELP (3,3).In addition, imported among the V-SEAT of virtual state, can estimate these edge shift.In addition, among the V-SEAT, do not comprise minimum run length (=3Tw), the oblique line part among the figure owing to do not consider virtual state, is therefore put on the identical value with S-SEAT to it in advance.By like this, use S-SEAT or V-SEAT, can estimate the shift amount of each bit code shape.During record, in order to make it, by determining that implementing examination by the pulse parameter of recording impulse width or position, edge institute structure writes, and can access the record condition that is suitable for PRML near 0.Equally, during regeneration,,, can access the regeneration condition that is suitable for PRML by determining to implement academic probation by regeneration equilibrium condition or focus offset in order to make it near 0.

Figure 53 makes the synoptic diagram of the flow process that appropriateization of condition of recording impulse try to write for explanation.The recording impulse parameter is defined as front and rear side along the table that is respectively 4 * 4 among the DVD-RAM.Above-mentioned SFP (s, m), ELP (s, m) corresponding with the recording impulse parameter.Edge sign indicating number shape for 4 * 4 forms is used S-SEAT of the present invention or V-SEAT, can estimate its shift amount.Simple order is that at first, the condition of change record pulse is carried out record in CD media.Corresponding S-SEAT or the value of V-SEAT are estimated in this sector of regenerating, and determine the parameter of recording impulse, make it near 0.Can learn by this example, since the recording impulse parameter be one to one as the S-SEAT value or the V-SEAT value of its evaluation of estimate, therefore, carry out recording/reproducing by once changing a plurality of recording impulse parameters, come simultaneously a plurality of recording impulse parameters appropriately to be changed side by side, can shorten examination and write the time.Specifically, begin to determine in turn the recording impulse parameter from the end, the processing time of the drive unit by 2 times of speed is about 30 seconds to 1 minute, and is relative therewith, by the present invention, implements to handle side by side, just can finish about 1 second to try to write.Under the situation that the examination of using S-SEAT is write, owing to can't measure SFP (3,3) and ELP (3,3), therefore, can make recording impulse parameter corresponding with it is SFP (3,3)=SFP (3,4).But said method only is the method that is similar to, and therefore, preferably uses the V-SEAT that can directly measure SFP (3,3) and ELP (3,3).

The example of Figure 54 for writing as the examination of using S-SEAT of the present invention or V-SEAT measured the position offset of recording impulse and the experimental result of the relation between each evaluation of estimate.Here, with the recording impulse of the 3Tw mark of DVD-RAM, do not change its width and experimentize from the starting position displacement.Figure 54 (a), Figure 54 (b), Figure 54 (c), Figure 54 (d) represent to use MLSE, the V-SEAT of PR (3,4,4,3) class respectively, use PR (a, b, c, d) S-SEAT of class and V-SEAT.Among the figure, SPF (3,3), SPF (6,3), ELP (3,3), ELP (6,3) have only been represented.Though not explanation in the top example, as previously mentioned, V-SEAT also can use for the PR (3,4,4,3) of fixed target level.Figure 54 (b) is an embodiment, and as shown in the figure, the pass between recording impulse displacement and each evaluation of estimate is a linear relationship, by making it near 0, just can easily realize the optimization of recording impulse parameter.But,, as previously mentioned, can't carry out the evaluation of SPF (3,3) and ELP (3,3) for MLSE, S-SEAT.

Figure 55 is the example of explanation as the academic probation of using S-SEAT of the present invention or V-SEAT, the experimental result of the relation between balanced recruitment and bit error rate and each evaluation of estimate.Here, and use PR (a, b, c, d) class changes the coefficient of 3 tap FIR wave filters of DVD-RAM standard equalizer as the PR class, changes recruitment.The measurement result of having represented MLSE, S-SEAT, V-SEAT among Figure 55 (a), Figure 55 (b), Figure 55 (c) respectively.Here, the PR class use the variable PR of target (a, b, c, d), the calculating correspondence of MLSE calculates target level by PR (3,4,4,3).As shown in the figure, make each index become minimum value, obtained the edge center condition of bit error rate substantially by selecting balanced recruitment.In the academic probation,,, determine balanced recruitment to make S-SEAT or V-SEAT minimum Yi Bian measure S-SEAT or V-SEAT on one side can change balanced recruitment.

Next, to the experimental result of the recording density of the 25Gb that is equivalent to blu-ray disc (Blu-ray Disc) as suitable example expression to RLL (1,7) coding.

The CD of being prepared as hereinbefore.Be on the substrate that the back-trough of orbit interval 0.34 μ m is constructed, the CD of lamination Write-once (write once) changing film.In the experiment, the DDU-1000 type evaluating apparatus that uses パ Le ス テ Star Network company to be produced.Optical source wavelength is 405nm, and the NA of object lens is 0.85.Use RLL (1,7) tone coded as separating, detection width Tw is 74.5nm.

Represented among Figure 56 use PR (1,2,2,1) and PR (a, b, c, d) during regeneration trial-production dish, the measured result of MLSE, S-SEAT and V-SEAT.The conventional balanced device that uses blu-ray disc is as balanced device, and recruitment is 6.0dB.To recording power is 3.2mW, and removing power is 0.35mW, rewrites 1 time in the groove track, and the regenerated signal quality of each mode is estimated.PR by the fixed target level (a, b, c, d) decoding is regenerated, and measures MLSE, regenerates by PR (1,2,2, the 1) decoding of target variable level, measures S-SEAT and V-SEAT.Each evaluation of estimate is the Gaussian distribution shape as shown in the figure, and transverse axis is as detection width, for ± Tw/2.Each evaluation of estimate is respectively MLSE=11.4%, S-SEAT=6.6%, V-SEAT=7.0%.Here, it is big that the MLSE value becomes, and distribute also from off-centring, is because owing to the conventional balanced device as the employed blu-ray disc of balanced device, makes and can't fully carry out equilibrium near the echo signal of PR (1,2,2,1) class.For example, if use the above FIR wave filter of 7 taps or 7 taps, the automatic equalization of enforcement use LSE (Least Square Error) method etc., just can improve distribution, but spinoff is also arranged, increased the recruitment of 2Tw signal, and increased the noise in high frequency band territory, therefore, this is not best regeneration condition.In addition, the PR that uses the target variable level (a, b, c, d) under the situation of class, PRML echo signal level changes along with regenerated signal, so the distribution of S-SEAT and V-SEAT is positioned at central authorities, disperses less.

Figure 57 is in above mensuration, estimates the result of the edge shift amount of the sign indicating number shape of distinguishing by the blank and the mark of front and back by MLSE, S-SEAT, V-SEAT.For MLSE that is subjected to run-length restriction and S-SEAT, can't estimate the front along displacement SFP (2,2) and back edge displacement ELP (2,2).In addition, imported among the V-SEAT of virtual state, can estimate these edge shift.In addition, among the V-SEAT, (value of=2Tw) part is identical with the value of S-SEAT not comprise minimum run length.By like this, use S-SEAT or V-SEAT, can estimate the shift amount of each bit code shape.During record, in order to make it, by determining that implementing examination by the pulse parameter of recording impulse width or position, edge institute structure writes, and can access the record condition that is suitable for PRML near 0.Equally, during regeneration,, in addition,,, can access the regeneration condition that is suitable for PRML by determining to implement academic probation by regeneration equilibrium condition or focus offset in order to make the value minimum of S-SEAT or V-SEAT in order to make it near 0.

Figure 58 is the example of expression as the academic probation of using S-SEAT of the present invention or V-SEAT, the experimental result of the relation between balanced recruitment and bit error rate and each evaluation of estimate.Here, and use PR (a, b, c, d) class is as the PR class.Because Viterbi decoder this point that V-SEAT can be applicable to absolute value system is by levies in kind, so represented among Figure 58 (b) that absolute value is the correlated results of Viterbi decoder.As shown in the figure, by selecting to make the balanced recruitment of V-SEAT minimum, can access the edge center condition of bit error rate substantially.In the academic probation,,, determine balanced recruitment to make the V-SEAT minimum Yi Bian measure V-SEAT on one side can change balanced recruitment.Though only represented the experimental result of V-SEAT here, for S-SEAT too, can determine that equilibrium condition makes its minimum.Certainly, S-SEAT also can be applicable to the Veterbi decoding of absolute value system.

[embodiment 5]

Optical disc apparatus

Figure 50 is the embodiment of the structure of explanation optical disc apparatus of the present invention.Make CD media 100 rotations by motor 160.During regeneration, 120 controls of laser power/impulse controller are flow through the electric current of the semiconductor laser 112 in the shaven head 110 and are produced laser 114, and being the specified light intensity of CPU140, laser 114 forms luminous point 101 by 111 optically focused of object lens on CD media 100.Reflected light 115 from this luminous point 101 is detected by photodetector 113 through object lens 111.Photodetector constitutes by being divided into a plurality of photodetectors.Read signal processing circuit 130 uses with shaven head 110 signals that detect, the information of regenerative recording on CD media 100.During record, laser power/impulse controller 120 is transformed to the pulse current of regulation with the record data of regulation, controls to make pulsed light from semiconductor laser 112 emissions.Decoding unit 10, target level unit 20 and signal evaluation unit 30 that structure regenerated signal of the present invention is estimated circuit are built in the read signal processing circuit 130.

Under the situation of estimating the regenerated signal quality, when examination is write according to the indication of CPU140, can playback of data, estimate S-SEAT or V-SEAT.For making this value minimum, to (1) focus bias, (2) equilibrium condition, the condition of (3) recording power and pulse is learnt, and just can make the recording optimization, and can make the action of optical disc apparatus stable.

Claims (29)

1. the evaluation method of a regenerated signal corresponding with the PRML mode, the echo signal level of above-mentioned PRML mode can change according to regenerated signal, it is characterized in that, comprising:

The correct bit-array that generation has been decoded and with the process of the error bit array of this bit-array edge shift 1 bit;

According to above-mentioned correct bit-array and error bit array,, generate the process of correct echo signal and false target signal respectively with reference to above-mentioned echo signal level;

2 powers of the signal level difference of each that calculates above-mentioned correct echo signal and above-mentioned regenerated signal in constantly and, calculate the process of correct Euclidean distance;

2 powers of the signal level difference of each that calculates above-mentioned false target signal and above-mentioned regenerated signal in constantly and, calculate the process of wrong Euclidean distance;

Poor as above-mentioned correct Euclidean distance and wrong Euclidean distance calculates the process of Euclidean distance difference;

According to above-mentioned echo signal level, calculate process corresponding to the average Euclidean distance of 1 bit shift code shape of above-mentioned echo signal level; After above-mentioned Euclidean distance difference is extracted above-mentioned average Euclidean distance out, divide by above-mentioned average Euclidean distance, calculate the process of normalization sequence error; And

Use above-mentioned normalization sequence error, estimate the process of above-mentioned regenerated signal.

2. the evaluation method of a regenerated signal, it is that the PRML mode of the coding more than 2 or 2 is corresponding with minimum run length, it is characterized in that, comprising:

Calculate the process of the brigade commander's littler echo signal level than minimum run length;

The correct bit-array that generation has been decoded and with the process of the error bit array of this bit-array edge shift 1 bit;

According to above-mentioned correct bit-array and error bit array,, generate the process of correct echo signal and false target signal respectively with reference to above-mentioned echo signal level;

2 powers of the signal level difference of each that calculates above-mentioned correct echo signal and above-mentioned regenerated signal in constantly and, calculate the process of correct Euclidean distance;

2 powers of the signal level difference of each that calculates above-mentioned false target signal and above-mentioned regenerated signal in constantly and, calculate the process of wrong Euclidean distance;

Poor as above-mentioned correct Euclidean distance and wrong Euclidean distance calculates the process of Euclidean distance difference:

According to above-mentioned echo signal level, calculate process with the corresponding average Euclidean distance of 1 bit shift code shape of above-mentioned echo signal level;

After above-mentioned Euclidean distance difference is extracted above-mentioned average Euclidean distance out, divide with above-mentioned average Euclidean distance, calculate the process of normalization sequence error; And

Use above-mentioned normalization sequence error, estimate the process of above-mentioned regenerated signal.

3. regenerated signal evaluation method as claimed in claim 2 is characterized in that,

Above-mentioned echo signal level can change according to regenerated signal.

4. regenerated signal evaluation method as claimed in claim 1 is characterized in that,

Evaluation of estimate σ by being calculated to (formula D-6) according to following (formula D-1) estimates above-mentioned regenerated signal.

[several 1]

${\mathrm{ED}}_B(\mathrm{pat}1,\mathrm{pat}2)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left(V_{t\arg \mathrm{et}}\right[\mathrm{pat}1\left[n\right]]-V_{t\arg \mathrm{et}}[\mathrm{pat}2\left[n\right]\left]\right)}^2$ (formula D-1)

(V _Target[B] is the echo signal level at bit-array B, pat[n] bit-array among the moment n of bit code shape " pat ", N representation class bit number)

[several 2]

$d_{\min }=\mathrm{Average}\left({\mathrm{ED}}_{\min }\right)$

$=\frac{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{\mathrm{ED}}_B\left({\mathrm{Pat}}_T\right[m],{\mathrm{Pat}}_F[m\left]\right)}{M}$ (formula D-2)

(M is the combination sum of 1 bit error code shape, Par _TAnd Pat _FRepresent correct bit code shape and error bit sign indicating number shape respectively)

[several 3]

$\mathrm{ED}\left(\mathrm{pat}\right)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left(V_{\mathrm{signal}}\right[t+n]-V_{t\arg \mathrm{et}}[\mathrm{pat}\left[n\right]\left]\right)}^2$ (formula D-3)

(V _SignalRegenerated signal level among [t] expression moment t, 2 value results among the t are bit code shape " pat " constantly)

[several 4]

D=Sign (Shift-Direction) * { (ED (Pat _F[m])-ED (Pat _T[m]))-d _Min(formula D-4) Sign (Shift-Direction)=-1 (Right-Edge-Shift)

+ 1 (Left-Edge-Shift) (formula D-5)

$\mathrm{\&sigma;}=\frac{\sqrt{\frac{\underset{p=1}{\overset{P}{\mathrm{\&Sigma;}}}{\left(D\right[p\left]\right)}^2}{P}}}{2d_{\min }}$ (formula D-6)

(Sign (Shift-Direction) expression makes 2 values Pat as a result _T1 bit mistake (edge shift) becomes Pat _FSituation under the edge shift direction, P represents the number of the bit code shape in computing interval of appointment)

5. regenerated signal evaluation method as claimed in claim 2 is characterized in that,

Evaluation of estimate σ by being calculated to (formula D-6) according to following (formula D-1) estimates above-mentioned regenerated signal.

[several 5]

${\mathrm{ED}}_B(\mathrm{pat}1,\mathrm{pat}2)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left(V_{t\arg \mathrm{et}}\right[\mathrm{pat}1\left[n\right]]-V_{t\arg \mathrm{et}}[\mathrm{pat}2\left[n\right]\left]\right)}^2$ (formula D-1)

(V _Target[B] expression is for the echo signal level of bit-array B, pat[n] bit-array among the moment n of expression bit code shape " pat ", N representation class bit number)

[several 6]

$d_{\min }=\mathrm{Average}\left({\mathrm{ED}}_{\min }\right)$

$=\frac{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{\mathrm{ED}}_B\left({\mathrm{Pat}}_T\right[m],{\mathrm{Pat}}_F[m\left]\right)}{M}$ (formula D-2)

(M is the combination sum of 1 bit error code shape, Pat _TAnd Pat _FRepresent correct bit code shape and error bit sign indicating number shape respectively)

[several 7]

$\mathrm{ED}\left(\mathrm{pat}\right)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left(V_{\mathrm{signal}}\right[t+n]-V_{t\arg \mathrm{et}}[\mathrm{pat}\left[n\right]\left]\right)}^2$ (formula D-3)

(V _SignalRegenerated signal level among [t] expression moment t, 2 value results among the t are bit code shape " pat " constantly)

[several 8]

D=Sign (Shift-Direction) * { (ED (Pat _F[m])-ED (Pat _T[m]))-d _Min(formula D-4) Sign (Shift-Direction)=-1 (Right-Edge-Shift)

+ 1 (Left-Edge-Shift) (formula D-5)

$\mathrm{\&sigma;}=\frac{\sqrt{\frac{\underset{p=1}{\overset{P}{\mathrm{\&Sigma;}}}{\left(D\right[p\left]\right)}^2}{P}}}{2d_{\min }}$ (formula D-6)

(Sign (Shiff-Direction) expression makes 2 values Pat as a result _T1 bit mistake (edge shift) becomes Pat _FSituation under the edge shift direction, P represents the number of the bit code shape in computing interval of appointment)

6. the recording method in the optical disc apparatus is characterized in that, comprising:

The parameter of change record pulse, the process that in CD media, writes down;

Regenerative recording data, the process of regenerated signal being estimated by the evaluation method described in the claim 1; And

According to above-mentioned evaluation, the process that the parameter of above-mentioned recording impulse is appropriately changed.

7. the recording method in the optical disc apparatus is characterized in that, comprising:

The parameter of change record pulse, the process that in CD media, writes down;

Regenerative recording data, the process of regenerated signal being estimated by the evaluation method described in the claim 2; And

According to above-mentioned evaluation, the process that the parameter of above-mentioned recording impulse is appropriately changed.

8. the signal regeneration method of an optical disc apparatus, it is used for through Viterbi decoder regenerated signal being decoded, and it is characterized in that, comprising:

Change the equilibrium condition of above-mentioned Viterbi decoder, the process of regenerated signal being estimated by the evaluation method described in the claim 1; And

Above-mentioned equilibrium condition is appropriately changed, made the process of above-mentioned evaluation maximum.

9. the signal regeneration method of an optical disc apparatus, it is used for through Viterbi decoder above-mentioned regenerated signal being decoded, and it is characterized in that, comprising:

Change the equilibrium condition of above-mentioned Viterbi decoder, the process of regenerated signal being estimated by the evaluation method described in the claim 2; And

Above-mentioned equilibrium condition is appropriately changed, made the process of above-mentioned evaluation maximum.

10. the regenerated signal evaluation method described in claim 2 is characterized in that,

Calculating the process of above-mentioned echo signal level, is to make the bit-array that includes the long bit of 1T effective, generates the process of echo signal.

11. an optical disc apparatus, it is equipped with the PRML regenerative loop, and above-mentioned PRML circuit can change the echo signal level according to regenerated signal, it is characterized in that, comprising:

The correct bit-array that generation has been decoded and with the device of the error bit array of this bit-array edge shift 1 bit;

According to above-mentioned correct bit-array and error bit array,, generate the device of correct echo signal and false target signal respectively with reference to above-mentioned echo signal level;

2 powers of the signal level difference of each that calculates above-mentioned correct echo signal and above-mentioned regenerated signal in constantly and, calculate the device of correct Euclidean distance; 2 powers of the signal level difference of each that calculates above-mentioned false target signal and above-mentioned regenerated signal in constantly and, calculate the device of wrong Euclidean distance;

Poor as above-mentioned correct Euclidean distance and wrong Euclidean distance calculates the device of Euclidean distance difference;

According to above-mentioned echo signal level, calculate device with the corresponding average Euclidean distance of 1 bit shift code shape of above-mentioned echo signal level;

After above-mentioned Euclidean distance difference is extracted above-mentioned average Euclidean distance out, divide with above-mentioned average Euclidean distance, calculate the device of normalization sequence error;

Have and to use above-mentioned normalization sequence error, the function that above-mentioned regenerated signal is estimated.

12. the optical disc apparatus described in claim 11 is characterized in that,

Above-mentioned PRML circuit, the device with the echo signal level that calculates the brigade commander littler than minimum run length;

Above-mentioned correct echo signal, false target signal generating apparatus, the echo signal level with reference to comprising the brigade commander's littler than above-mentioned minimum run length echo signal level generates correct echo signal, false target signal respectively.

13. the evaluation method of a regenerated signal corresponding with the PRML mode, above-mentioned PRML mode can change the echo signal level according to regenerated signal, it is characterized in that, comprising:

The correct bit-array that generation has been decoded and with the process of the error bit array of this bit-array edge shift 1 bit;

According to above-mentioned correct bit-array and error bit array,, generate the process of correct echo signal and false target signal respectively with reference to above-mentioned echo signal level;

2 powers of the signal level difference of each that calculates above-mentioned correct echo signal and above-mentioned regenerated signal in constantly and, calculate the process of correct Euclidean distance;

2 powers of the signal level difference of each that calculates above-mentioned false target signal and above-mentioned regenerated signal in constantly and, calculate the process of wrong Euclidean distance;

Poor as above-mentioned correct Euclidean distance and wrong Euclidean distance calculates the process of Euclidean distance difference;

According to above-mentioned echo signal level, calculate process with the corresponding average Euclidean distance of 1 bit shift code shape of above-mentioned echo signal level; After above-mentioned Euclidean distance difference is extracted above-mentioned average Euclidean distance out, divide by above-mentioned average Euclidean distance, calculate the process of normalization sequence error; And

Use above-mentioned normalization sequence error, estimate the process of above-mentioned regenerated signal.

14. the evaluation method of a regenerated signal, it is that the PRML mode of the coding more than 2 or 2 is corresponding with minimum run length, it is characterized in that, comprising:

Calculate the process of the brigade commander's littler echo signal level than minimum run length; The correct bit-array that generation has been decoded and with the process of the error bit array of this bit-array edge shift 1 bit;

According to above-mentioned correct bit-array and error bit array,, generate the process of correct echo signal and false target signal respectively with reference to above-mentioned echo signal level;

2 powers of the signal level difference of each that calculates above-mentioned correct echo signal and above-mentioned regenerated signal in constantly and, calculate the process of correct Euclidean distance; 2 powers of the signal level difference of each that calculates above-mentioned false target signal and above-mentioned regenerated signal in constantly and, calculate the process of wrong Euclidean distance; Poor as above-mentioned correct Euclidean distance and wrong Euclidean distance calculates the process of Euclidean distance difference; According to above-mentioned echo signal level, calculate process with the corresponding average Euclidean distance of 1 bit shift code shape of above-mentioned echo signal level; After above-mentioned Euclidean distance difference is extracted above-mentioned average Euclidean distance out, divide with above-mentioned average Euclidean distance, calculate the process of normalization sequence error; And

Use above-mentioned normalization sequence error, estimate the process of above-mentioned regenerated signal.

15. regenerated signal evaluation method as claimed in claim 13 is characterized in that,

Evaluation of estimate σ by being calculated to (formula D-6) according to following (formula D-1) estimates above-mentioned regenerated signal.

[several 9]

${\mathrm{ED}}_B(\mathrm{pat}1,\mathrm{pat}2)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left(V_{t\arg \mathrm{et}}\right[\mathrm{pat}1\left[n\right]]-V_{t\arg \mathrm{et}}[\mathrm{pat}2\left[n\right]\left]\right)}^2$ (formula D-1)

(V _Target[B] is the echo signal level for bit-array B, pat[n] bit-array among the moment n of bit code shape " pat ", N representation class bit number)

[several 10]

$d_{\min }=\mathrm{Average}\left({\mathrm{ED}}_{\min }\right)$

$=\frac{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{\mathrm{ED}}_B\left({\mathrm{Pat}}_T\right[m],{\mathrm{Pat}}_F[m\left]\right)}{M}$ (formula D-2)

(M is the combination sum of 1 bit error code shape, Par _TAnd Pat _FRepresent correct bit code shape and error bit sign indicating number shape respectively)

[several 11]

$\mathrm{ED}\left(\mathrm{pat}\right)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left(V_{\mathrm{signal}}\right[t+n]-V_{t\arg \mathrm{et}}[\mathrm{pat}\left[n\right]\left]\right)}^2$ (formula D-3)

(V _SignalRegenerated signal level among [t] expression moment t, 2 value results among the t are bit code shape " pat " constantly)

[several 12]

D=Sign (Shift-Direction) * { (ED (Pat _F[m])-ED (Pat _T[m]))-d _Min(formula D-4) Sign (Shift-Direction)=-1 (Right-Edge-Shift)

+ 1 (Left-Edge-Shift) (formula D-5)

$\mathrm{\&sigma;}=\frac{\sqrt{\frac{\underset{p=1}{\overset{P}{\mathrm{\&Sigma;}}}{\left(D\right[p\left]\right)}^2}{P}}}{{2d}_{\min }}$ (formula D-6)

(Sign (Shift-Direction) expression makes 2 values Pat as a result _T1 bit mistake (edge shift) becomes Pat _FSituation under the edge shift direction, P represents the number of the bit code shape in computing interval of appointment)

16. regenerated signal evaluation method as claimed in claim 14 is characterized in that,

Evaluation of estimate σ by being calculated to (formula D-6) according to following (formula D-1) estimates above-mentioned regenerated signal.

[several 13]

${\mathrm{ED}}_B(\mathrm{pat}1,\mathrm{pat}2)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left(V_{t\arg \mathrm{et}}\right[\mathrm{pat}1\left[n\right]]-V_{t\arg \mathrm{et}}[\mathrm{pat}2\left[n\right]\left]\right)}^2$ (formula D-1)

(V _Target[B] is the echo signal level for bit-array B, pat[n] bit-array among the moment n of bit code shape " pat ", N representation class bit number)

[several 14]

$d_{\min }=\mathrm{Average}\left({\mathrm{ED}}_{\min }\right)$

$=\frac{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{\mathrm{ED}}_B\left({\mathrm{Pat}}_T\right[m],{\mathrm{Pat}}_F[m\left]\right)}{M}$ (formula D-2)

(M is the combination sum of 1 bit error code shape, Par _TAnd Pat _FRepresent correct bit code shape and error bit sign indicating number shape respectively)

[several 15]

$\mathrm{ED}\left(\mathrm{pat}\right)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left(V_{\mathrm{signal}}\right[t+n]-V_{t\arg \mathrm{et}}[\mathrm{pat}\left[n\right]\left]\right)}^2$ (formula D-3)

(V _SignalRegenerated signal level among [t] expression moment t, 2 value results among the t are " pat " constantly)

[several 16]

D=Sign (Shift-Direction) * { (ED (Pat _F[m])-ED (Pat _T[m]))-d _Min(formula D-4) Sign (Shift-Direction)=-1 (Right-Edge-Shift)

+ 1 (Left-Edge-Shift) (formula D-5)

$\mathrm{\&sigma;}=\frac{\sqrt{\frac{\underset{p=1}{\overset{P}{\mathrm{\&Sigma;}}}{\left(D\right[p\left]\right)}^2}{P}}}{2d_{\min }}$ (formula D-6)

(Sign (Shift-Direction) expression makes 2 values Pat as a result _T1 bit mistake (edge shift) becomes Pat _FSituation under the edge shift direction, P represents the number of the bit code shape in computing interval of appointment)

17. a regenerated signal evaluation method is characterized in that,

In claim 15 or claim 16, replace (formula D-1) and (formula D-3), respectively by following (formula D-7) and (formula D-8), with Euclidean distance as each in constantly both difference absolute value with calculate.

[several 17]

${\mathrm{ED}}_B(\mathrm{pat}1,\mathrm{pat}2)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}\left|V_{t\arg \mathrm{et}}\right[\mathrm{pat}1\left[n\right]]-V_{t\arg \mathrm{et}}[\mathrm{pat}2\left[n\right]\left]\right|$ (formula D-7)

$\mathrm{ED}\left(\mathrm{pat}\right)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}\left|V_{\mathrm{signal}}\right[t+n]-V_{t\arg \mathrm{et}}[\mathrm{pat}\left[n\right]\left]\right|$ (formula D-8)

18. a regenerated signal evaluation method is characterized in that,

In claim 15 or claim 16, replace (formula D-2), benchmark Euclidean distance d _MinAs the value in this yard shape, use following (formula D-9).

[several 18]

d _Min=ED _B(Pat _T[m], Pat _F[m]) (formula D-9)

Here, ED _B(formula D-7) following by (formula D-1) calculates

[several 19]

${\mathrm{ED}}_B(\mathrm{pat}1,\mathrm{pat}2)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}\left|V_{t\arg \mathrm{et}}\right[\mathrm{pat}1\left[n\right]]-V_{t\arg \mathrm{et}}[\mathrm{pat}2\left[n\right]\left]\right|$ (formula D-7)

19. a regenerated signal evaluation method is characterized in that,

In claim 15 or claim 16, replace (formula D-4), establish to about the evaluation of estimate of edge shift when being respectively DL, DR, under the situation that this DL and DR can both calculate, use following (formula D-10) represented average shift amount, as the shift amount D at 1 edge being had in mind.

[several 20]

$D=\frac{(\mathrm{DL}+\mathrm{DR})}{2}$ (formula D-10)

20. a regenerated signal evaluation method is characterized in that,

In claim 15 or claim 16, replace (formula D-4), establish to about the evaluation of estimate of edge shift when being respectively DL, DR, under the situation that this DL and DR can both calculate, use the less side's of following (formula D-11) represented such absolute value shift amount, as the shift amount D at 1 edge being had in mind.

[several 21]

D=DL if|DL|≤| DR| (formula D-11)

D＝DR if|DL|＞|DR|

21. a regenerated signal evaluation method is characterized in that,

In claim 15 or claim 16, replace (formula D-4), establish to about the evaluation of estimate of edge shift when being respectively DL, DR, under the situation that this DL and DR can both calculate, use the bigger side's of following (formula D-12) represented such absolute value shift amount, as the shift amount D at 1 edge being had in mind.

[several 22]

D=DR if|DL|≤| DR| (formula D-12)

D＝DL if|DL|＞|DR|

22. an optical disc apparatus, it is equipped with the PRML regenerative loop, and above-mentioned PRML circuit can change the echo signal level according to regenerated signal, it is characterized in that, comprising:

The correct bit-array that generation has been decoded and with the device of the error bit array of this bit-array edge shift 1 bit; According to above-mentioned correct bit-array and error bit array,, generate the device of correct echo signal and false target signal respectively with reference to above-mentioned echo signal level; The absolute value of the signal level difference of each that calculates above-mentioned correct echo signal and above-mentioned regenerated signal in constantly and, calculate the device of correct Euclidean distance; The absolute value of the signal level difference of each that calculates above-mentioned false target signal and above-mentioned regenerated signal in constantly and, calculate the device of wrong Euclidean distance; Poor as above-mentioned correct Euclidean distance and wrong Euclidean distance calculates the device of Euclidean distance difference; According to above-mentioned echo signal level, calculate device with the corresponding average Euclidean distance of 1 bit shift code shape of above-mentioned echo signal level; And

After above-mentioned Euclidean distance difference is extracted above-mentioned average Euclidean distance out, divide by above-mentioned average Euclidean distance, calculate the device of normalization sequence error,

Have and to use above-mentioned normalization sequence error, the function that above-mentioned regenerated signal is estimated.

23. an IC chip, it is made of Veterbi decoding circuit with waveform equalizer and target level table and signal evaluation circuit, it is characterized in that,

Above-mentioned signal evaluation circuit comprises a yard shape selected cell, the 1st and the 2nd target level computing unit, sequence error evaluation unit;

Above-mentioned sign indicating number shape selected cell, the 2 value results that above-mentioned Veterbi decoding circuit is exported, corresponding to 1 bit mistake, bit number * 2-1 of preservation class judges whether to calculate with a sign indicating number shape into evaluation of estimate, is being judged as under the situation of evaluation of estimate calculating with sign indicating number shape, it is sent to above-mentioned the 1st target level computing unit as correct bit code shape, simultaneously, generate the edge error code shape of 1 bit that misplaced, send to above-mentioned the 2nd target level computing unit as error bit sign indicating number shape;

Above-mentioned the 1st target level computing unit, with reference to above-mentioned target level table, output and the correct corresponding echo signal level of bit code shape, above-mentioned the 2nd target level computing unit, with reference to above-mentioned target level table, export the echo signal level corresponding with error bit sign indicating number shape;

Above-mentioned sequence error evaluation unit uses the above-mentioned the 1st and the output of the output of the 2nd target level computing unit and above-mentioned waveform equalizer etc., according to following (formula D-1) to (formula D-6), and any among output evaluation of estimate D or the σ.

[several 23]

${\mathrm{ED}}_B(\mathrm{pat}1,\mathrm{pat}2)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left(V_{t\arg \mathrm{et}}\right[\mathrm{pat}1\left[n\right]]-V_{t\arg \mathrm{et}}[\mathrm{pat}2\left[n\right]\left]\right)}^2$

(formula D-1)

(V _Target[B] is the echo signal level for bit-array B, pat[n] bit-array among the moment n of bit code shape " pat ", N representation class bit number)

[several 24]

$d_{\min }=\mathrm{Average}\left({\mathrm{ED}}_{\min }\right)$

$=\frac{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{\mathrm{ED}}_B\left({\mathrm{Pat}}_T\right[m],{\mathrm{Pat}}_F[m\left]\right)}{M}$ (formula D-2)

(M is the combination sum of 1 bit error code shape, Pat _TAnd Pat _FRepresent correct bit code shape and error bit sign indicating number shape respectively)

[several 25]

$\mathrm{ED}\left(\mathrm{pat}\right)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left(V_{\mathrm{signal}}\right[t+n]-V_{t\arg \mathrm{et}}[\mathrm{pat}\left[n\right]\left]\right)}^2$ (formula D-3)

(V _SignalRegenerated signal level among [t] expression moment t, 2 value results among the t are bit code shape " pat " constantly)

[several 26]

D=Sign (Shift-Direction) * { (ED (Pat _F[m])-ED (Pat _T[m]))-d _Min(formula D-4) Sign (Shift-Direction)=-1 (Right-Edge-Shift)

+ 1 (Left-Edge-Shift) (formula D-5)

$\mathrm{\&sigma;}=\frac{\sqrt{\frac{\underset{p=1}{\overset{P}{\mathrm{\&Sigma;}}}{\left(D\right[p\left]\right)}^2}{P}}}{{2d}_{\min }}$ (formula D-6)

(Sign (Shift-Direction) expression makes 2 values Pat as a result _T1 bit mistake (edge shift) becomes Pat _FSituation under the edge shift direction, P represents the number of the bit code shape in computing interval of appointment)

24. an IC chip, it is characterized in that by Veterbi decoding circuit with waveform equalizer and target level table and signal evaluation circuit structure,

Above-mentioned signal evaluation circuit comprises a yard shape selected cell, the 1st and the 2nd target level computing unit, sequence error evaluation unit and virtual target level calculation unit;

Above-mentioned virtual target level calculation unit does not rely on run-length restriction, uses each tabular value of above-mentioned target level table, calculates the echo signal level to all bit-array, preserves as the virtual target water-glass;

Above-mentioned sign indicating number shape selected cell, the 2 value results that above-mentioned Veterbi decoding circuit is exported, corresponding to 1 bit mistake, bit number * 2-1 of preservation class judges whether to calculate with a sign indicating number shape into evaluation of estimate, is being judged as under the situation of evaluation of estimate calculating with sign indicating number shape, it is sent to above-mentioned the 1st target level computing unit as correct bit code shape, simultaneously, generate the edge error code shape of 1 bit that misplaced, send to above-mentioned the 2nd target level computing unit as error bit sign indicating number shape;

Above-mentioned the 1st target level computing unit, with reference to above-mentioned virtual target water-glass, output is corresponding to the echo signal level of correct bit code shape, above-mentioned the 2nd target level computing unit, with reference to above-mentioned virtual target water-glass, output is corresponding to the echo signal level of error bit sign indicating number shape;

Above-mentioned sequence error evaluation unit uses the above-mentioned the 1st and the output of the output of the 2nd target level computing unit and above-mentioned waveform equalizer etc., according to following (formula D-1) to (formula D-6), and any among output evaluation of estimate D or the σ.

[several 27]

${\mathrm{ED}}_B(\mathrm{pat}1,\mathrm{pat}2)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left(V_{t\arg \mathrm{et}}\right[\mathrm{pat}1\left[n\right]]-V_{t\arg \mathrm{et}}[\mathrm{pat}2\left[n\right]\left]\right)}^2$ (formula D-1)

(V _Target[B] is the echo signal level for bit-array B, pat[n] bit-array among the moment n of bit code shape " pat ", N representation class bit number)

[several 28]

$d_{\min }=\mathrm{Average}\left({\mathrm{ED}}_{\min }\right)$

$=\frac{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{\mathrm{ED}}_B\left({\mathrm{Pat}}_T\right[m],{\mathrm{Pat}}_F[m\left]\right)}{M}$ (formula D-2)

(M is the combination sum of 1 bit error code shape, Pat _TAnd Pat _FRepresent correct bit code shape and error bit sign indicating number shape respectively)

[several 29]

$\mathrm{ED}\left(\mathrm{pat}\right)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left(V_{\mathrm{signal}}\right[t+n]-V_{t\arg \mathrm{et}}[\mathrm{pat}\left[n\right]\left]\right)}^2$ (formula D-3)

(V _SignalRegenerated signal level among [t] expression moment t, 2 value results among the t are bit code shape " pat " constantly)

[several 30]

D=Sign (Shift-Direction) * { (ED (Pat _F[m])-ED (Pat _T[m]))-d _Min(formula D-4) Sign (Shift-Direction)=-1 (Right-Edge-Shift)

+ 1 (Left-Edge-Shift) (formula D-5)

$\mathrm{\&sigma;}=\frac{\sqrt{\frac{\underset{p=1}{\overset{P}{\mathrm{\&Sigma;}}}{\left(D\right[p\left]\right)}^2}{P}}}{{2d}_{\min }}$ (formula D-6)

(Sign (Shift-Direction) expression makes 2 values Pat as a result _T1 bit mistake (edge shift) becomes Pat _FSituation under the edge shift direction, P represents the number of the bit code shape in computing interval of appointment)

25. an IC circuit is characterized in that,

In claim 23 or claim 24, replace (formula D-1) and (formula D-3), respectively by following (formula D-7) and (formula D-8), with Euclidean distance as each in constantly both difference absolute value with calculate.

[several 31]

${\mathrm{ED}}_B(\mathrm{pat}1,\mathrm{pat}2)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}\left|V_{t\arg \mathrm{et}}\right[\mathrm{pat}1\left[n\right]]-V_{t\arg \mathrm{et}}[\mathrm{pat}2\left[n\right]\left]\right|$ (formula D-7)

$\mathrm{ED}\left(\mathrm{pat}\right)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}\left|V_{\mathrm{signal}}\right[t+n]-V_{t\arg \mathrm{et}}[\mathrm{pat}\left[n\right]\left]\right|$ (formula D-8)

26. an IC circuit is characterized in that,

In claim 23 or claim 24, replace (formula D-2), benchmark Euclidean distance d _MinAs the value in this yard shape, use following (formula D-9).

[several 32]

d _Min=ED _B(Pat _T[m], Pat _F[m]) (formula D-9)

Here, ED _BCalculate by (formula D-1) or following (formula D-7)

[several 33]

${\mathrm{ED}}_B(\mathrm{pat}1,\mathrm{pat}2)=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}\left|V_{t\arg \mathrm{et}}\right[\mathrm{pat}1\left[n\right]]-V_{t\arg \mathrm{et}}[\mathrm{pat}2\left[n\right]\left]\right|$ (formula D-7)

27. an IC circuit is characterized in that,

In claim 23 or claim 24, shift amount D as 1 edge being had in mind, replace (formula D-4), establish to about the evaluation of estimate of edge shift when being respectively DL, DR, under the situation that this DL and DR can both calculate, use following (formula D-10) represented average shift amount.

[several 34]

$D=\frac{(\mathrm{DL}+\mathrm{DR})}{2}$ (formula D-10)

28. an IC circuit is characterized in that,

In claim 23 or claim 24, shift amount D as 1 edge being had in mind, replace (formula D-4), establish to about the evaluation of estimate of edge shift when being respectively DL, DR, under the situation that this DL and DR can both calculate, use the less side's of following (formula D-11) represented such absolute value shift amount.

[several 35]

D=DL if|DL|≤| DR| (formula D-11)

D＝DR if|DL|＞|DR|

29. a regenerated signal evaluation method is characterized in that,

In claim 23 or claim 24, shift amount D as 1 edge being had in mind, replace (formula D-4), establish to about the evaluation of estimate of edge shift when being respectively DL, DR, under the situation that this DL and DR can both calculate, use the bigger side's of following (formula D-12) represented such absolute value shift amount.

[several 36]

D=DR if|DL|≤| DR| (formula D-12)

D＝DL if|DL|＞|DR|

Images