JP4048641B2 - Playback apparatus and playback method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a playback apparatus and playback method for playing back digital data recorded on a recording medium such as an optical disk.
[0002]
[Prior art]
As optical recording media for recording digital data, compact discs (CD), mini discs (MD), and 130 mm and 90 mm optical disc cartridges according to the International Organization for Standardization (ISO) standards have been commercialized. In recent years, a DVD standard optical disc having a recording capacity about 7 times that of a CD has appeared, and it is attracting attention as a recording medium for moving image data whose data amount is increasing in the future.
[0003]
Conventionally, a CD or the like employs a bit-by-bit decoding method in which a read analog signal is determined to be “1” if it is larger than a predetermined level and “0” if it is smaller, but the recording density of the DVD standard is greatly improved. With such a recording medium, it is difficult to reproduce digital data with high reliability.
[0004]
Therefore, in recent years, as a method for reproducing digital data, a PRML method combining a partial response method (PR method or PR) and a maximum likelihood decoding method (ML method or ML) using a Viterbi decoding method has attracted attention. .
[0005]
[Problems to be solved by the invention]
Techniques using the PRML system for signal reproduction of optical recording media include, for example, JP-A-8-172366, JP-A-8-221910, or JP-A-8-263934. In Japanese Patent Laid-Open No. 8-172366, when a Viterbi decoder obtains a decoded data string using a square error between a received sample value and a predicted value as a branch metric, it is approximated by a polygonal line by a linear function to reduce the circuit scale. Is disclosed. However, although this method reduces the circuit scale, it cannot cope with an equalization method having an intermediate value with good equalization characteristics, and is not suitable for reproducing high-density data.
[0006]
In Japanese Patent Laid-Open No. 8-221910, when obtaining a sample value from a read signal, the amplitude is limited so that a predicted value when Viterbi decoding is performed can be reduced. Even if asymmetry occurs in the read signal, the branch metric value can be forcibly set to 0, so that the Viterbi decoding performance can be prevented from deteriorating. However, even with this method, an error from the predicted value near the intermediate value such as the median value remains, so that an equalization method with an intermediate value with good equalization characteristics cannot provide much effect.
[0007]
In Japanese Patent Laid-Open No. 8-263934, a sample value smaller than a predetermined value is extracted from sample values of a reproduction signal, and the phase of a sampling clock pulse for A / D conversion is corrected based on the polarity and inclination of the sample value. The level is automatically corrected to the optimum state. However, in this method, since it is necessary to provide a clock pulse whose phase can be adjusted exclusively for sampling, the circuit scale for digital processing increases.
[0008]
Thus, in the present invention, branch metric processing can be performed at high speed even when an equalization method with good equalization characteristics and a large amount of intermediate values is adopted, and further, the influence of asymmetry specific to the reproduction signal of the optical recording medium is affected. It is an object of the present invention to provide a reproducing apparatus and a reproducing method that are strong and have high decoding performance. It is another object of the present invention to simplify the circuit configuration, operate with low power consumption, and to supply a highly reliable reproducing apparatus that can cope with high recording density at low cost.
[0009]
[Means for Solving the Problems]
For this reason, in the present invention, two prediction values corresponding to the median value (zero cross) among the prediction values at the time of maximum likelihood decoding are provided, and the prediction value and the sample are obtained when an asymmetric read signal is obtained by asymmetry. By reducing the error from the value, high-speed and highly reliable digital data can be reproduced. That is, the playback apparatus of the present invention includes a partial response equalization means for performing waveform equalization using a partial response on a read signal obtained from a recording medium on which digital data is recorded and converting the read signal into a sample value series, and a continuous in the sample value series. Based on the difference value of sample values to be selected, one of two predicted values corresponding to the median value of the sample value series is selected, and the sample value series is subjected to maximum likelihood decoding using the selected predicted value to reproduce digital data. Maximum likelihood decoding means. Further, the reproduction method of the present invention includes a partial response equalization step of performing waveform equalization by a partial response on a read signal obtained from a recording medium on which digital data is recorded and converting the read signal into a sample value sequence, and a continuous in the sample value sequence. Based on the difference value of sample values to be selected, one of two predicted values corresponding to the median value of the sample value series is selected, and the sample value series is subjected to maximum likelihood decoding using the selected predicted value to reproduce digital data. And a maximum likelihood decoding step.
[0010]
By providing two prediction values corresponding to the median value of the sample value series, the read signal is strongly affected by asymmetry, and the median value is biased positively or negatively. Even when the sample value at the changing timing becomes asymmetric, it can be grasped at high speed and more accurately that it is the median value. For this reason, branch metric calculation and path metric calculation at the time of maximum likelihood decoding can be performed at high speed. Further, since the median is evaluated with high accuracy, highly accurate digital data can be decoded. Therefore, it is possible to provide a playback device and a playback method with high speed and good decoding performance.
[0011]
The two predicted values corresponding to these median values are preferably generated from the sample value series by a predetermined algorithm. By enabling the predicted value to be updated with the sample value series, it is possible to absorb the influence on the sample value due to the condition of the recording medium being read or the environment in which the playback device is installed. Maximum likelihood decoding can be performed with a predicted value with little error from the value. Therefore, it is possible to provide a playback device and a playback method with higher speed and higher accuracy. To generate a predicted value from a sample value series, for example, the median (zero cross) is determined by taking the first-order difference and the second-order difference, and the predicted value is calculated from the sample value at that time or an appropriate average value thereof. Can be sought. Further, by providing prediction values with positive and negative first-order differences, even when an asymmetric sample value sequence is obtained by asymmetry, it can be efficiently decoded. Further, the predicted value may be updated from the median value and the predicted sample value, or an appropriate average value thereof.
[0012]
Furthermore, by generating not only the predicted value corresponding to the median value of the sample value series but also other predicted values corresponding to the waveform equalized sample value series, the decoding efficiency can be further improved. Therefore, in addition to the first predicted value generating means or the first predicted value generating step for generating a predicted value near the median, the second predicted value generating means or the second for generating other predicted values by a predetermined algorithm. It is desirable to provide a predicted value generation step.
[0013]
When there is asymmetry in the read signal, the analysis of the inventors of the present application has revealed that the sample values obtained by reading the equalized waveform at the detection points are dispersed and deviated from the predicted values. It is desirable to generate an appropriate predicted value shifted from the theoretical value. Also, it is considered to change the equalization coefficient in the partial response equalization means in accordance with asymmetry. Even in this case, the variance of the sample values is reduced, but the sample values are shifted from the predicted values. Therefore, in order to decode a read signal having asymmetry, it is very effective to generate each prediction value in consideration of the sample value series.
[0014]
One method for obtaining a predicted value is to obtain an average value of sample values dispersed upward from the median value and an average value of sample values dispersed downward from the sample value series, and to obtain these average values. In other words, other predicted values are generated by adding or subtracting the difference of the theoretical predicted value from the average value. Another method is to generate sample values that are consecutive multiple times in the sample value series as the highest predicted value or the lowest predicted value, and calculate a preset coefficient for these predicted values. In other words, another predicted value between the predicted value corresponding to the median value is generated.
[0015]
Furthermore, it is desirable that the partial response equalization means can delay-correct the timing for obtaining the sample value series. As a result, the level of the sample value can be automatically corrected to an optimum state without controlling a clock signal for performing A / D conversion for generating a digitized sample value series. For this reason, a common clock signal can be supplied and moved to the conversion means, the partial response equalization means, and the maximum likelihood decoding means. Therefore, the circuit scale of the reproducing apparatus can be reduced and the power consumption can be reduced. The read signal may be A / D converted and then input to the partial response equalization means, or the sample value series output from the partial response equalization means may be A / D converted and digitized.
[0016]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of an optical disc apparatus 1 according to an embodiment. The processing in the optical disc apparatus 1 of this example is performed along the arrow in the block diagram, and the disc data recorded on the optical disc medium 2 is read and reproduced. First, in the optical disk apparatus 1 of this example, the optical data is recorded on the optical disk medium 2 by irradiating the optical disk medium 2 with a laser beam by the optical pickup 3 and is obtained as an analog read signal φg. The analog signal is amplified. The amplified analog signal is shaped by a waveform shaping circuit 5 having functions of a low frequency pass filter (LPF) and an equalizer (EQ) that performs waveform shaping. Thereafter, the signal is converted into a digital signal by an analog / digital (A / D) converter 6. Further, the read signal φi obtained by digitizing the read signal φg is processed by the partial response (PR) equalization circuit 10 to generate a sample value series φs having a waveform equalized with a predetermined partial response. Then, the sample value series φs is Viterbi-decoded by the maximum likelihood decoding circuit 20, and the reproduction signal φo of the digital data recorded on the optical disc medium 2 is decoded.
[0017]
The optical disc apparatus 1 of this example includes a PLL that outputs a clock signal φc in addition to the optical pickup 3, the preamplifier 4, the waveform shaping circuit 5, the A / D conversion circuit 6, the PR equalization circuit 10, and the maximum likelihood decoding circuit 20. A circuit 7 is provided. In the PLL circuit 7, a signal having a frequency synchronized with the bit period of the signal read from the optical disc medium 2 is obtained by referring to the analog signal output from the waveform shaping circuit 5 and from which high-frequency noise is cut. Therefore, by using the signal output from the PLL circuit 7 as a clock signal, a clock signal suitable for the A / D conversion circuit 6, the waveform equalization circuit 10, and the maximum likelihood decoding circuit 20 can be obtained.
[0018]
The PR equalization circuit 10 of this example includes a 3-tap equalization filter (transversal filter, Finite Impulse Response (FIR) filter) 11 that performs waveform equalization, and a delay correction circuit 12. As shown in FIG. 2, the equalization filter 11 is obtained in advance according to three delay elements 15 connected in series, an adder 16 for adding digital signals before and after these delay elements 15 and an equalization method. And a multiplier 17 for multiplying the weight coefficient (equalization coefficient). The delay element 15 is a delay element that can secure a delay time D of 1 bit for the clock signal φc. Further, in the equalization circuit 11 of this example, equalization coefficients of the multipliers 17 are set so that sample values can be obtained by the PR (1221) method.
[0019]
FIG. 3 shows the evaluation of the bit error rate (BER) by the 8/16 code and various PRML systems adopted for the DVD standard optical disk. As shown in this figure, it can be seen that the PR (1221) method has the smallest error rate, and the PR (1221) method becomes more advantageous as the S / N ratio increases. As described above, the PR (1221) method is suitable for reproducing digital data of an optical disc of the DVD standard. This is considered because the frequency characteristic of the PR (1221) method approximates the MTF (Modulation Transfer Function) that is the frequency characteristic of the optical system of the DVD standard. Therefore, it is desirable to adopt the PR (1221) system for the reproducing apparatus of the optical disk apparatus that handles the DVD standard optical disk, and it is possible to reduce an equalization error that occurs when performing waveform equalization and to suppress noise.
[0020]
FIG. 4 shows the relationship between the tap number of the equalization filter and the bit error rate (BER) when the PR (1221) method is adopted. The number of taps is the number of delay elements 15 of the equalization filter 11. When the PR (1221) method is adopted, the BER hardly changes even if the sample value series is lengthened by increasing the number of taps. Therefore, the equalization filter 11 of this example employs a three-tap configuration that is the minimum number of taps for obtaining a PR (1221) method sample value sequence. By employing an equalization filter with a small number of taps, the circuit scale of the PR equalization circuit 10 can be reduced, the size of the reproducing chip can be reduced, and the power consumption can be suppressed.
[0021]
The delay correction circuit 12 provided in the PR equalization circuit 10 of the present example uses a multiplication circuit to add several minutes to the input clock signal φc in addition to the first delay element D having a delay time of 1 bit. A second delay element capable of creating a one-bit period delay is provided. Also, the signal digitization period in the A / D conversion circuit 6 is set to be shorter than the delay period D due to the 1-bit period, and the digital signal φi can be obtained with a fraction of a 1-bit period. For this reason, it is possible to digitally adjust the timing for obtaining the sample value series φs by adjusting the delay period of the second delay element. Therefore, the PR equalizer circuit 10 can control the sampling timing without controlling the phase of the clock signal φc and adjusting the A / D conversion timing. For this reason, it is possible to process the A / D converter, the A / D converter 6, the PR equalization circuit 10, and the maximum likelihood decoding circuit 20 with a digital circuit that operates with the same clock signal φc. Therefore, in the optical disc apparatus 1 of this example, an analog circuit for adjusting the phase of the clock signal is unnecessary, so that the circuit scale of the reproduction system of the optical disc apparatus 1 can be made very small.
[0022]
The method of delay correction in the PR equalization circuit 10 is not limited to the above. A sample value series φs that can be obtained at an appropriate timing can be obtained by using the data that has been A / D converted with the clock having a shifted timing and performing the time shift calculation shown below.
[0023]
Y (k) = α (U (k) −Y (k−1)) + U (k−1) (1)
However, U (k) is a value actually sampled at the k-th timing, and Y (k) is an appropriate timing with no delay Δt due to A / D conversion or the like with respect to the 1-bit transfer time To. A sample value that would have been obtained. Α is expressed as follows.
[0024]
α = (1−Δt / T0) / (1 + Δt / T0) (2)
By performing this calculation, the delay Δt associated with the sampling of the A / D converter, which has been known in advance, the sample value U (k) obtained at the delayed timing, and further obtained at the previous timing. Based on the sample values Y (k−1) and U (k−1), the sample value Y (k) that would be obtained at an appropriate timing can be output from the PR equalization circuit 10.
[0025]
The maximum likelihood decoding circuit 20 reproduces the digital signal based on the sample value series φs obtained by the PR equalization circuit 10 in this way. The maximum likelihood decoding circuit 20 evaluates the sample value of the sample value series φs and the predicted value φr by the least square method, and evaluates the sample value evaluated by the branch metric calculation circuit 21 using a path metric. And a path memory 23 for decoding digital data based on the evaluated sample value series.
[0026]
In the optical disc apparatus 1 of the present example, two predicted values φrm, which are intermediate (median) among the predicted values φr referred to by the branch metric calculation circuit 21, are prepared, and these φrm + and φrm− are converted into a sample value series φs. A first predicted value generation unit 8 is generated. Then, one of these two predicted values φrm + and φrm− is selected according to the trend of the sample value series φs or the trellis diagram, and used in branch metric calculation and path metric calculation.
[0027]
FIG. 5 shows an example of the sample value series φs obtained by the PR equalization circuit 10. In the DVD standard, 8/16 modulation that modulates 8-bit data into 16-bit data is adopted, and the minimum pit length is 3 and the maximum pit length is 11. Therefore, when 8 / 16-modulated digital data is equalized by the PR (1221) method, each of the predicted values φr (φr0, φr1, φr3, φr5, and φr6) of (0, 1, 3, 5, 6) is obtained. A sample value series φs in which corresponding values are combined is obtained.
[0028]
For example, when the digital data shown in FIG. 5A is equalized by the PR (1221) method, the sample value series φs shown in FIG. 5B is obtained. This figure shows an ideal sample value sequence, but the actually obtained sample value sequence is shifted in various directions from each predicted value φr. The main cause is the asymmetry peculiar to the optical disc in addition to noise. Asymmetry appears due to the characteristics of the optical pickup 3 or other reproduction system due to the pit length formed on the disc being deviated from the specified length due to excessive or insufficient laser power during mastering. With asymmetry, the reproduced signal is asymmetric because the median is biased positively or negatively.
[0029]
In the optical disc apparatus 1 of this example, the deviation of the sample value due to the deviation of the sampling timing can be corrected by the delay correction circuit 12 of the PR equalization circuit 10 described above. When the sample value sequence φs is asymmetric due to asymmetry, the maximum likelihood is obtained by preparing two central prediction values (3 in this example) φrm in the positive and negative directions (φrm + and φrm−). In the decoding circuit 20, even an asymmetric sample value sequence φs can be accurately evaluated in a short time.
[0030]
The value at the detection point of the waveform equalized data, that is, the shift amount (error) due to the asymmetry of the sample value may be due to factors unique to each optical disc medium or factors unique to the optical disc apparatus 1. Therefore, it is desirable to employ the predicted value φr that meets the conditions for reproducing digital data. However, even if a plurality of predicted values are prepared, it is difficult to evaluate which value is suitable as the predicted value. Especially, when there is asymmetry, the central sample value fluctuates greatly, so it is possible to automatically set the predicted value. Have difficulty. On the other hand, in the optical disc apparatus 1 of this example, since two predicted values φrm + and φrm− are prepared in the plus direction and the minus direction, the sample values are likely to converge. Therefore, even when multiple prediction values are prepared for the median value, it is possible to determine a prediction value suitable for decoding relatively easily by evaluating the sample values separately in the positive and negative directions. Can do.
[0031]
The predicted value may be prepared in advance, but the automatically generated value can be made more suitable for the evaluation of the sample value series φs obtained at that time. Furthermore, as described above, by preparing two predicted values φrm + and φrm− in the plus direction and the minus direction, the values are likely to converge, so that the predicted values can be automatically generated more easily. Then, by using a predicted value close to the automatically generated sample value series φs, accurate evaluation can be performed in a short time as described above. It is easy to limit the upper and lower sample values using a limiter as described above. Therefore, by automatically setting the median value, the distance between the sample value series and the predicted value becomes very close, and the error rate can be greatly improved.
[0032]
Several algorithms for automatically generating the central predicted value φrm that can be adopted by the first predicted value generation unit 8 are conceivable, and FIG. 5 shows an example thereof. The median value of the sample value series φs can be regarded as the median point, and its position (timing) is 1 as shown in FIGS. 5B and 5C, including the plus direction and the minus direction. This can be determined by taking the floor difference and the second floor difference. That is, a sample value when the first-order difference is plus and the second-order difference is 0 is generated as a median predicted value φrm + in the plus direction, and the sample value when the first-order difference is minus and the second-order difference is 0 Can be generated and output as the predicted value φrm− of the median value in the minus direction. Therefore, the predicted value generation unit 8 can obtain appropriate predicted values φrm + and φrm− by averaging the sample values at the respective timings to some extent according to such an algorithm. Then, the predicted values can be optimized by updating the generated predicted values φrm + and φrm− with the predicted values in the predicted value memory 24 of the maximum likelihood decoding circuit 20 at an appropriate timing.
[0033]
The algorithm is not limited to this. For example, an average of sample values determined as a median value in the plus direction or the minus direction by a branch metric calculation circuit 21 described later is taken, and each predicted value is updated at an appropriate timing. can do.
[0034]
Furthermore, the optical disc apparatus 1 of the present example includes a second predicted value generation unit 9, and uses other intermediate predicted values φrm + and φrm− generated by the first predicted value generation unit 8. The predicted value φr is generated from the sample value series φs. For example, in the case of PR (1221), the predicted values φr0, φr1, φr5, and φr6 are obtained by the second predicted value generation unit 9, stored in the predicted value memory 24, and can be used by the branch metric circuit 21. . Therefore, in the optical disc apparatus 1 of this example, as shown in the flowchart of FIG. 6, the step 51 for generating the sample value series φs by waveform equalizing the read signal φg read from the optical disc 2, and the sample value series φs In addition to step 52 for maximum likelihood decoding, first, a step (first predicted value generation step) 55 for obtaining predicted values φrm + and φrm− corresponding to the intermediate value from the sample value series φs, and another predicted value φr0 , Φr1, φr5, and φr6 (second predicted value generation step) 56. In step 52 for maximum likelihood decoding, the prediction value generated based on the sample value series φs in steps 55 and 56 is used, so that the decoding efficiency is very high. Further, in this example, since two prediction values corresponding to the intermediate value are prepared, even when the read signal φg has asymmetry, the maximum likelihood decoding circuit 20 has a high decoding efficiency and a low error rate. Can provide. If there is an asymmetry, it is very effective in improving the decoding capability to generate other predicted values φr0, φr1, φr5 and φr6 based on the sample value sequence φs in step 56. Has been found by the inventors' research.
[0035]
FIG. 7 schematically shows a voltage level of a waveform obtained by performing PR equalization on a signal when there is no asymmetry and when there is asymmetry. First, FIG. 7A shows voltage hiss and gram of a waveform obtained by PR equalizing the read signal φg without asymmetry, and 0 is a theoretical voltage level (predicted value) when equivalent to PR (1221). It can be seen that sample values with small dispersion centered on 1, 3, 5 and 6 are obtained. On the other hand, FIG. 7B is a voltage histogram of a waveform obtained by PR equalizing the read signal φg including the positive asymmetry. FIG. 7C is a voltage histogram of a waveform obtained by PR equalizing the read signal φg including negative asymmetry. As can be seen from these figures, when there is asymmetry, the variance of sample values increases, and the average of each sample value also shifts up and down from the theoretical predicted value. Therefore, when such sample value series φs is maximum likelihood decoded with a theoretical prediction value, the error rate also increases.
[0036]
On the other hand, the probability that the median can be determined is greatly improved by moving the median, that is, the predicted value corresponding to the voltage level 3, up and down according to a sample value with positive or negative asymmetry. And by discriminating the median with high accuracy, the upper and lower sample values can be discriminated very easily, so that the decoding capability is greatly improved. Further, in the optical disc reproducing apparatus 1 of this example, other predicted values are generated according to the sample values, so that the decoding capability is greatly improved and the error rate is also improved.
[0037]
Further, in the sample value series φs shown in FIG. 7, if there is asymmetry, not only the average value of each sample value is shifted, but also the variance is large, so the load on the branch metric circuit 21 becomes large. For this reason, in the PR equalization circuit 10, it is considered to reduce the dispersion by changing the equalization coefficient to that for asymmetry. The voltage histogram of the sample value series φs obtained as a result is schematically shown in FIG.
[0038]
As in FIG. 7, FIG. 8A is a voltage histogram of a sample value series obtained by PR equalizing the read signal φg without asymmetry, and FIG. 8B shows the read signal φg with positive asymmetry. It is a voltage histogram of the sample value series equalized by PR. FIG. 8C is a voltage histogram of a sample value series obtained by PR equalizing the read signal φg having negative asymmetry. As can be seen from these figures, the variance of each sample value can be reduced by switching the equalization coefficient of the multiplier 17 for obtaining the sample value series φs in the PR equalization circuit 10 to the one scheduled for asymmetry. However, their average value shifts up or down from the theoretical prediction. Therefore, it can be seen that it is very useful to use the predicted value φr generated corresponding to the sample value sequence φs in order to perform maximum likelihood decoding of the sample value sequence φs having such a shifted value.
[0039]
Several methods for generating the predicted value φr corresponding to sample values other than the intermediate value are conceivable. One of them is a method of actively using the predicted values φrm + and φrm− corresponding to the intermediate value obtained by the first predicted value generation circuit 8, and is a sample value sequence φs equivalent by PR (1221). Predicted values φr0, φr1, φr5, and φr6 corresponding to the respective sample values can be obtained by the following equations.
[0040]
φr0 = Avl-0.5
φr1 = Avl + 0.5
φr5 = Avh−0.5
φr6 = Avh + 0.5 (3)
However, Avl is the average of sample values below the median value of the sample value series φs (low voltage level), Avh is the average of sample values above the median value of the sample value series (high voltage level) is there.
[0041]
That is, in this algorithm, the average of the sample values above φrm is considered to be the average of the sample values originally having a voltage level of 5 or 6, and the average value is obtained to adjust the difference between the theoretical prediction values. Thus, predicted values φr5 and φr6 corresponding to the respective sample values are generated. Predicted values φr0 and φr1 can be generated similarly.
[0042]
The algorithm for obtaining the predicted value uses an intermediate predicted value as a threshold, and has an advantage that the algorithm is simple and the circuit scale is small. However, it is pointed out that the determination method with a threshold value is generally vulnerable to DC fluctuation. On the other hand, it is also possible to generate a predicted value of a sample value other than the intermediate value by using the PR-equalized waveform characteristics as described below.
[0043]
First, FIG. 9 shows some of the waveforms subjected to PR equalization (in this example, PR (1221) is shown as described above). FIG. 9A shows a 3T waveform (T is a channel bit length) corresponding to the shortest bit of 8/16 modulation employed in a DVD, and sample values (ideal) are only 1, 3 and 5. I don't take it. On the other hand, in the 4T waveform shown in FIG. 9B, the voltage levels of 0, 1, 3, 5 and 6 are taken, and in the 5T waveform shown in FIG. 9C, the highest level (the highest level) is taken. Signal) 6 continues, and the lowest level (lowest signal) 0 continues. Further, in the 6T waveform shown in FIG. 9 (d), three voltage levels 6 continue, three voltage levels 0 also continue, in the 7T waveform shown in FIG. 9 (e), four voltage levels 6 continue, Four voltage levels 0 continue. In 8/16 modulation, the longest code is 11T, but as can be seen from the above, the number of consecutive voltage levels 6 and 0 is further increased.
[0044]
Focusing on such a sample value series φs, if three or more differential values of consecutive sample values are zero or close to sufficiently zero, these sample values have a theoretical value (voltage level) of 6 or It can be determined that it belongs to 0. Therefore, the average value of these consecutive sample values can be used to obtain the predicted value φr6 of level 6 or the predicted value φr0 of level 0, and the predicted value can be obtained by comparing these values with the predicted value φrm of the intermediate value. φr6 and φr0 can be generated.
[0045]
When the predicted values φr6 and φr0 are obtained, the remaining predicted values φr5 and φr1 can be obtained by the following equations.
[0046]
φr5 = φrm + A (φr6-φrm)
φr1 = φr0 + B (φrm−φr0) (4)
However, A and B are coefficients.
[0047]
In PR (1221), the coefficient A can be about 2/3 and the coefficient B can be about 1/3. However, if there is asymmetry, it can be changed according to the shift of the sample value, and the error rate can be further improved by adjusting the coefficients A and B.
[0048]
Thus, not only the intermediate value predicted values φrm + and φrm− are generated from the sample value series φs, but also the error rate can be improved by generating predicted values φr corresponding to other sample values, The effect is remarkable when there is asymmetry.
[0049]
When each predicted value φr is generated in this way, the maximum likelihood decoding circuit 20 evaluates the sample value series φs based on the predicted value φr. First, the branch metric calculation circuit 21 performs a square calculation of the difference between the sample value of the sample value series φs and the predicted value, and evaluates by the least square error method. Next, the squared signal is input to the path metric calculation circuit 22. The path metric calculation circuit 22 is an ACS (Add Compare Select) circuit including an adder, a comparator, and a selector. The path metric calculation circuit 22 adds the path metric value and the square error signal by the adder, and has two types. The addition output signal is selected as a comparison value by a comparator, and a small value is selected by a selector and held as the current path metric value. The selection information at that time is output to and held in the path memory 23, and a value corresponding to the maximum likelihood path is output as reproduction data φo from the last stage of the path memory 23.
[0050]
As described above, the optical disc apparatus 1 of the present example equalizes and outputs a signal read by the PR (1221) method suitable for reproducing digital data stored on a DVD standard disc, and further outputs a central value as a predicted value. By providing two values, the sample value series equalized and output by the PR (1221) method is evaluated with six predicted values instead of the five conventional predicted values. Accordingly, the PR (1221) type trellis diagram of the maximum likelihood decoding circuit 20 of this example has six states with five values as shown in FIG. 10, but the five values are decoded into six. Since the two predicted values for evaluating the median value can be updated based on the sample value series, the calculation time in the branch metric calculation circuit and the path metric calculation circuit can be shortened, and the accuracy can be increased. Furthermore, it is possible to reduce the computation time and the circuit scale by removing state transitions that cannot be obtained by 8/16 modulation adopted in the DVD standard, and it is possible to obtain high-speed and highly reliable reproduction data. .
[0051]
In this example, the PR (1221) playback apparatus is described as an example in which the equalization method is suitable for reproducing an optical disc, particularly an optical disc according to the DVD standard. However, a reproduction circuit employing another equalization method is described. Of course, the present invention can also be applied to the reproduction method. In addition, the reproducing apparatus and the reproducing method of the present invention are not limited to DVDs, but can be applied to other optical disks such as MDs and apparatuses that reproduce data from data recording media such as magnetic disks.
[0052]
【The invention's effect】
As described above, in the playback apparatus and playback method employing the PRML method of the present invention, the read signal is strongly influenced by asymmetry by providing two prediction values for evaluating the median value of the sample value series. Even when the median is biased positively or negatively and becomes asymmetrical when it changes positively or negatively, the sample value series can be evaluated appropriately and in a short time. Further, these predicted values are generated from the sample value series by a predetermined algorithm, and high reliability can be obtained even when the environment of the recording medium or reproducing apparatus to be read is different.
[0053]
Furthermore, in addition to the median value, the error rate can be improved by generating predicted values corresponding to other sample values based on the sample value series. In particular, when there is asymmetry, the inventors of the present application have found that the PR-equalized sample value shifts up and down, and by generating a predicted value corresponding to each sample value, It is possible to provide a reproducing apparatus and a reproducing method having a high reproduction capability and a low error rate even for a certain read signal.
[0054]
In addition, when performing PR equalization, the sampling timing can be adjusted without changing the A / D conversion timing in order to obtain a digitized sample value sequence by delay-correcting the sampling timing of the sample value sequence. . Therefore, the circuit scale of the reproducing apparatus can be reduced and the power consumption can be reduced.
[0055]
Equalization in a playback apparatus and playback method employing a PR (1221) system whose frequency characteristics are similar to the MTF characteristics of an optical system employed in an optical disc apparatus of the DVD standard as an equalization system, or another equalization system Error and noise can be suppressed. Therefore, it is possible to provide a high-speed and highly reliable reproducing apparatus and reproducing method suitable for reproducing a signal of an optical disc having a high recording density.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a schematic configuration of an optical disc apparatus according to an embodiment of the present invention.
2 is a diagram showing a schematic configuration of an equalization filter of a PR equalization circuit of the optical disc apparatus shown in FIG. 1. FIG.
FIG. 3 is a graph showing error rates when 8 / 16-modulated data is reproduced by various PRML systems;
FIG. 4 is a graph showing the relationship between the number of taps of the PR (1221) type equalization filter and the error rate.
FIG. 5 is a diagram illustrating an example of a sample value series of a PR (1221) method.
6 is a flowchart showing an outline of a reproduction process in the optical disc apparatus shown in FIG.
FIG. 7 is a voltage histogram showing how sample values vary depending on the presence or absence of asymmetry.
FIG. 8 is a voltage histogram showing sample values when the equalization coefficient of the PR equalization circuit is switched to a coefficient corresponding to asymmetry.
FIG. 9 is a diagram showing some examples of PR (1221) equalized sample value series;
FIG. 10 is a trellis diagram of the PR (1221) system.
[Explanation of symbols]
1 Optical disk device
2 Optical disc
3 Optical pickup
4 Preamplifier
5 Waveform shaping circuit
6 A / D conversion circuit
7 Clock source
8 First predicted value generator
9 Second predicted value generator
10 PR equalization circuit
11 Equalization filter
12 Delay correction circuit
20 Maximum likelihood decoding circuit
21 Branch metric arithmetic circuit
22 Path metric calculation circuit
23 path memory
24 Predicted value memory

Claims (17)

Partial response equalization means for performing waveform equalization by a partial response of a read signal obtained from a recording medium on which digital data is recorded and converting the read signal into a sample value series;
Based on a difference value between consecutive sample values in the sample value series, one of two predicted values corresponding to a median value of the sample value series is selected, and the sample value series is selected using the selected predicted value And a maximum likelihood decoding means for reproducing the digital data.   2. The reproducing apparatus according to claim 1, further comprising first predicted value generation means for generating two predicted values corresponding to the median value from the sample value series by a predetermined algorithm.   The first predicted value generation means according to claim 2, wherein the first-order difference and the second-order difference of the sample value series are obtained to generate a predicted value with a positive first-order difference and a negative predicted value. Playback device.   3. The reproduction according to claim 2, further comprising second predicted value generation means for generating another predicted value by a predetermined algorithm based on the predicted value corresponding to the median value generated by the first predicted value generating means. apparatus.   5. The second predicted value generation means according to claim 4, wherein, in the sample value series, an average value of sample values distributed upward with respect to the median value and an average value of sample values distributed downward are respectively obtained. A reproduction apparatus characterized by generating other predicted values by adding or subtracting a difference from the average value of theoretical predicted values to the obtained average value.   5. The second predicted value generation unit according to claim 4, wherein the second predicted value generation means generates a sample value that is consecutive multiple times as the highest predicted value or the lowest predicted value in the sample value series, and stores these predicted values in advance. A playback apparatus, wherein a set coefficient is calculated to generate another predicted value between the predicted value corresponding to the median value.   2. The reproducing apparatus according to claim 1, wherein the partial response equalization means includes means for delay correcting the timing for obtaining the sample value series.   8. The reproducing apparatus according to claim 7, further comprising a clock signal supply unit that supplies a common clock signal to the conversion unit, the partial response equalization unit, and the maximum likelihood decoding unit.   2. The reproducing apparatus according to claim 1, wherein the recording medium is an optical recording medium. A partial response equalization step in which a read signal obtained from a recording medium on which digital data is recorded is subjected to waveform equalization by a partial response and converted into a sample value series;
Based on a difference value between consecutive sample values in the sample value series, one of two predicted values corresponding to a median value of the sample value series is selected, and the sample value series is selected using the selected predicted value And a maximum likelihood decoding step of reproducing digital data by maximum likelihood decoding.   11. The reproduction method according to claim 10, further comprising a first predicted value generation step of generating two predicted values corresponding to the median value from the sample value series by a predetermined algorithm.   12. The first predicted value generation step according to claim 11, wherein a first-order difference and a second-order difference of the sample value series are obtained, and a predicted value with a positive first-order difference and a negative predicted value are generated. How to play.   12. The reproduction according to claim 11, further comprising a second predicted value generation step of generating another predicted value by a predetermined algorithm based on the predicted value corresponding to the median value generated by the first predicted value generation means. Method.   14. The second predicted value generation step according to claim 13, wherein, in the sample value series, an average value of sample values dispersed upward with respect to the median value and an average value of sample values dispersed downward are respectively obtained. A reproduction method characterized by generating other predicted values by adding or subtracting a difference from the average value of theoretical predicted values to the obtained average value.   14. The second predicted value generation step according to claim 13, wherein a sample value that is continuous a plurality of times in the sample value series is generated as the highest-order predicted value or the lowest-order predicted value, and these predicted values are stored in advance. A reproduction method, wherein a set coefficient is calculated to generate another predicted value between the predicted value corresponding to the median value.   11. The reproducing method according to claim 10, wherein, in the partial response equalization step, the timing for obtaining the sample value series can be delay-corrected.   11. The reproducing method according to claim 10, wherein the recording medium is an optical recording medium.

Images