JPH0219550B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to an error correction prevention method in a PCM recording and reproducing apparatus. PCM recording/playback devices, which convert analog audio signals into digital signals and record them on tapes or video disks and play them back, have recently become a hot topic. Since the PCM signal occupies a wide frequency band, VTRs and video disks are used, but in such high-density recording and reproducing systems, many signal dropouts occur due to scratches and dust on the recording medium. Errors in digital signals due to dropouts include random errors and burst errors, and various code correction methods have been announced for each. Japan Electronics Industry Association Technical File STC-007 “Consumer PCM
According to "Encoder/Decoder", in a PCM signal recording/playback system using a VTR, burst errors are randomized by interleaving operation, and erroneous signals are corrected using adjacent code correction for random errors. The method has been adopted as a standard. The outline of a PCM signal recording and reproducing system using such a VTR will be briefly explained with reference to FIG. In Figure 1, audio signals are input to input terminals 1a and 1b (generally, stereo signals are input).
Lch, Rch) are converted into digital sampling signal words (hereinafter referred to as information data) in the AD conversion circuit 2, and an error correction word (hereinafter referred to as correction data) generated from the information data is added. AD conversion circuit 2 includes a low pass filter,
It includes a sample hold circuit, a data addition circuit for L/R time division circuit correction, and the like. After this data is interleaved in memory circuit 3,
Generated by TV composite signal generation circuit 4
The TV synchronization signal and the CRC etc. generated in the error detection word (hereinafter referred to as CRC) generation circuit 6 are mixed in the mixing circuit 5, and are converted into a standard TV signal to the VTR 7.
recorded in The output of the VTR 7 is as shown in Figure 2 a and b. Figure a shows the data arrangement within one horizontal synchronization period (1H), in which three pieces of information data each for Lch and Rch and two pieces of correction data P, Q and CRC are arranged, forming a data block. are doing. Figure b shows the data array within one field.
The data block shown in figure a is 245 in the 245H section.
Arranged in pieces. Although not shown in FIG. 2B, control signal blocks consisting of various control words are arranged in the 1H period preceding the 245 data blocks. Now, returning to Fig. 1, the video signal output from the VTR 7 is synchronously separated by the comparator circuit 8.
Data is separated. Since the data output from the comparator circuit 8 is interleaved data, it is de-interleaved using the memory circuit 9, and furthermore, the erroneous data is corrected by the correction circuit 10. After that, it is converted into the original analog audio signal by the DA conversion circuit 11 and sent to the output terminal 12.
a, 12b. Actually, the DA conversion circuit 11 includes an L/R time division circuit, a degritcher circuit, an LPF, and the like. The above is an overview of a PCM signal recording/playback system using a VTR.
De-interleaving will be briefly explained. Figure 3a shows the case where data is arranged in the same order without interleaving.When recording and playing back in this state, if there is a small random error, the erroneous data can be corrected and restored. However, when a burst error occurs, most of the data during that time is continuously lost, making it impossible to correct it. Therefore, as shown in Fig. 3b, one block of data, for example, W ₁ to W ₆ , P ₁ , Q ₁
Interleaving is not arranging them in the same H, but rearranging them in separate H's. FIG. 3b shows an example in which each piece of data is arranged 3H apart, which is generally referred to as interleaving with an interleaving interval of D=3H, and according to the standard, D=16H. De-interleaving means performing the opposite operation of interleaving to return to the original arrangement. By interleaving and recording PCM signals in this manner, even if a burst error occurs during playback, it can be treated as an apparently random error, and a high correction rate can be achieved. The PCM signal (video signal) shown in Figure 2b, interleaved and recorded as described above, is de-interleaved and corrected during playback, but the vertical synchronization signal V shown in Figure 2b is disturbed. Or, due to omissions, etc., the input data to the memory circuit 9 in FIG. 1, ie, the data 245H shown in FIG. 2b, may become incorrect. For example, if we consider the case where the 1H section indicated by the arrow F in Figure 3b falls through the cracks, the data input to the memory circuit 9 will be arranged in the memory as shown in Figure 4. It is thought that The input data to the correction circuit 10 is now the data read from the memory circuit 9 along a diagonal line G from top to bottom and from top left to bottom right in FIG. Fourth
If we take the word name in column L ₁ at the left end of the diagram and use it as the block name, there is no problem with correction up to block W ₁ , but the data combination between blocks W ₇ to _{W 121} is different from the original data combination. If correction work is performed during this period (21H), erroneous corrections may occur and large clicking noises may occur. Also, even if there is no error data and no correction work during this period, the order of the deinterleaved data will be out of order (for example, W ₄₃ , W ₄₄ , W ₄₅ , W ₄₆ , W ₄₇ , W ₅₄ ,
W ₄₉ , W ₅₀ , W ₅₁ , W ₅₂ , W ₅₃ , W ₆₀ , ...), the sound becomes abnormal during these 15 hours. Table 1 shows word shift when the above de-interleaving error occurs.

[Table] In the example of D=3H, the first 3H (3 data blocks)
In D ₁ , 1 data Q is shifted, and in the next 3H (3 data blocks), in D ₂ , 2 data, that is, P and Q are shifted, and 3 data are shifted in sequence.
data, 4 data, 5 data, 6 data, and 7 data. However, from the perspective of the remaining data, the deviations of data 6 and 7 are considered to be deviations of data 2 and 1, and this is the first The numbers and types are written in parentheses in the table. By the way, in the above standard, since D = 16H, there is a possibility of error correction in the section D ₁ to D ₇ = 16H x 7 = 112H, and furthermore, the section in which an error occurs due to a shift in the order of data is D ₃ ~D ₇ = 16H x 5 = 80H. Furthermore, this phenomenon also occurs when playing a tape that has been lightly edited (TV signals are appropriately connected and edited frame by frame). SUMMARY OF THE INVENTION An object of the present invention is to provide an error correction prevention method that minimizes the probability of occurrence of an error correction. In the present invention, when data deviation (misalignment) is found during correction work for a certain block, if the number of error data detected in the next block is greater than or equal to a predetermined number, the correction work is not performed. It is characterized by the fact that it was done as follows. In explaining the content of the present invention, adjacent code correction will be briefly explained using the above standard as an example. Information data W ₁ , W ₂ , ... W ₆ (1 word
14Bit), the correction data P and Q are as follows: P= ₆ 〓 ⁱ⁼¹ W _i = W ₁ W ₂ W ₃ W ₄ W ₅ W ₆ Q= ₆ 〓 ⁱ⁼¹ T ^7-i W _i =T ⁶ W ₁ T ⁵ W ₂ T ⁴ W ₃ T ³ W ₄ T ² W ₅
TW ₆ (here, T is the auxiliary matrix of the generator polynomial 1 + x ⁸ + x ¹⁴ ), and rewriting the previous equation, W ₁ W ₂ W ₃ W ₄ W ₅ W ₆ P=0 - T ⁶ W ₁ T ⁵ W ₂ T ⁴ W ₃ T ³ W ₂ T ² W ₅ TW ₆
Q=0 -. Now, suppose that the i-th and j-th data are mistaken and become W^ _i = W _i E _i W _j = W^ _j E _j - (E _i and E _j are error patterns), then the solution of the equation (syndrome ) do not become 0, but become S ₁ and S ₂ as shown below, respectively. S ₁ = E _i E _j S ₂ = T ^7-i E _i T ^7-j E _j . These two equations become E _j = (IT ^ij ) ^-1 (S ₁ T ^i-7 S ₂ ) E _i = S ₁ E _j , and the respective error patterns E _i and E _j can be calculated, and the equation can be Using this, W _i and W _j are calculated. By the way, the solution (syndrome) on the left side of the equation
When calculating, incorrect data was also added, but
Do not add incorrect data or add “0” (the result is the same as not adding it),
If the respective solutions are S _P and S _Q , then S _P = W _i W _j S _Q = T ^7-i W _i T ^7-j W _j , and from these two equations, W _j = (IT ^ij ) ^-1 (S _P T ^i-7 S _Q ) W _i =S _P W _j is calculated, and W _i and W _j can be directly calculated by the same calculation as the above-mentioned calculation of E _i and E _j . Therefore, in the following explanation, a case will be described in which a correction operation is performed in a manner that does not add erroneous data. Generally, if the number of error data detected by CRC in one block is "0", the syndrome P _P =0, S _Q =0 should be established. However, even if the number of erroneous data is "0", let's consider what kind of symptoms will appear if data deviation (misalignment) occurs due to the reasons described above. Assuming that there is a one-data shift and data W _i comes in, S _P =W _i W _i ′≠0 S _Q =T ^7-i (W _i W _i ′)≠0. Now, if we define the third syndrome as S _PQ = S _P T ^k-7 S _Q , then k is 1, 2, 3,...6
When k=i, S _PQ = S _P T ^i-7 S _Q = (W _i W _i ′) T ^i-7・T ^7-i (W _i W _i ′) = 0. , Which data is off (wrong)?
It shows that. Therefore, the value of i can be determined, and the above
W _i can be corrected. Correction data P,
Unlike the information data W _i , Q does not actually need to be corrected, but P and Q can also be corrected using the same method as W _i . When the correction data P is off, the following symptoms are obtained: syndrome S _P = PP'≠0 〃 S _Q = 0 〃 S _PQ = PP'≠0, and the correction data Q is off. In this case, the following symptoms are obtained: syndrome S _P = 0 〃 S _Q = QQ'≠0 〃 S _PQ = T ^k-7 (QQ')≠0. In this way, when the correction data P or Q is deviated, one of the syndromes S _P and S _Q becomes 0, and the remaining two syndromes are not 0. Therefore, it is possible to determine whether the correction data P is out of place (incorrect) or the correction _{data Q} is out of alignment (incorrect). By performing the calculation, the original correction data P or Q is calculated and corrected. Generally, these corrections are called syndrome corrections, and conversely, if syndrome corrections are made during correction work for a certain block, it can be determined that data deviations (misalignments) have occurred within that block. Can be done. The intervals D ₁ and D ₇ in Table 1 above are the intervals where the syndrome just described may be corrected. Therefore, in this section,
If a data shift is detected due to the occurrence of syndrome correction work, and the next block correction work is to correct two data errors, it is possible to prevent erroneous corrections by stopping this work. . In addition, in the section D ₂ to _{D 6} in Table 1, there is a deviation (deviation) of 2 or more data, so 1
Even if the number of error data detected by CRC in a block is “0”, three syndromes
S _P , S _Q , and S _PQ do not each become “0”. Therefore, once a data shift is detected, if the number of error data detected by CRC in one block is "0" or "1", and if there are three syndromes, S _P , S _Q ,
Erroneous correction can be prevented by stopping the correction of two data errors until at least one of S _PQ becomes "0". FIG. 5 is a block diagram of an embodiment of the present invention, where 9 and 10 are the memory circuits and correction circuits shown in FIG. 1, and 13 is a flag (for example, 14 is an error flag detection circuit that determines whether the data is erroneous by detecting the flag from the data output from the memory circuit 9; 15 is a syndrome The calculation means 16 is a control circuit that stores the outputs of the error flag detection circuit 14 and the syndrome calculation means 15 and controls the correction operation of the correction circuit 10 based on these outputs. As explained in detail above, according to the present invention,
This makes it possible to prevent de-interleaving operation errors and erroneous corrections on tapes that have been simply edited. In addition, in the above standard, interleaving length D=
Since it is 16H, D ₂ to _{D 6} in Table 1 above = 16H×
5 = 80H, so even if it is not incorrectly corrected, de-
Since the interleaved data is out of order, no clicking sound is produced, but it is played back as an abnormal sound. Therefore, PCM during the correction suspension period
A flag is set in the signal information data to indicate that there is a data shift, and this flag can be used to correct abnormal sounds such as pre-hold or zero level. It is possible to reduce the level. In a PCM signal recording and reproducing system that records and reproduces digitalized analog audio signals, the generation of click noise is extremely harmful. Click noise due to erroneous correction is one of them, and according to the present invention, the probability of occurrence of erroneous correction can be greatly reduced. Therefore, it has become possible to reproduce stable sound even when using a tape or tape recorder that is in poor condition, or when playing a tape that has been simply edited.

[Brief explanation of drawings]

Figure 1 is a block diagram of a PCM recording/playback device, Figure 2 is a recording signal format output from the PCM recording/playback device, and a is
Figure 3b is a data array diagram within the 1H interval, Figure 3b is the data array diagram within the 1V interval, Figure 3a is the data array diagram where the data is arranged in the same order without interleaving, Figure 3b is the data array diagram with the data interleaved. (D=
3H), FIG. 4 is a data array diagram when the 1H section in FIG. 3b is missing, and FIG. 5 is a block diagram of an embodiment of the present invention. 1a, 1b...Input terminal, 2...AD conversion circuit, 3...Memory circuit, 4...TV composite signal generation circuit, 5...Mixing circuit, 6...Error detection word generation circuit, 7...VTR, 8... Comparator circuit, 9... Memory circuit, 10... Correction circuit, 11... DA conversion circuit, 12a, 12b...
...Output terminal, 13...CRC error detection circuit, 14
. . . error flag detection circuit, 15 . . . syndrome calculation means, 16 . . . control circuit.

Claims (1)

[Claims] 1. A plurality of sampled signal words and an error correction word generated from the sampled signal words are treated as one data block, and the sampled signal words and error correction words are interleaved into a data string. After recording with an error detection word added and performing error detection using the error detection word during playback,
The data block is deinterleaved and reproduced, and the PCM performs correction using the error correction word for each reproduced data block.
In the system recording/reproducing apparatus, a counting means for counting the number of error words detected by the error detection word in one reproduced data block, and an error detection means for detecting errors in the correction process using the error correction word. If the counted value of the counting means is 0 in the reproduced first data block and the error detection means determines that an error exists, an abnormality occurs in the combination of data within one data block. Even if the number of error words detected by the error detection word in the next second data block is the number of words that can be corrected by the correction operation, the correction operation is stopped. An error correction prevention method characterized by the following. 2 Even if the number of error words detected by the error detection word in the data block following the second data block is the number of words that can be corrected by the correction operation, the correction operation for that data block will continue. 2. The error correction prevention method according to claim 1, wherein the error correction prevention method is configured such that the above operation is continued until the number of error words in subsequent data blocks becomes zero.