KR830002525B1 - Pcm signal transmition method - Google Patents

Abstract

This procedure transmite a sequence of N digital information words assembled in a transmission block. The words are separated into respective sequences of odd and even information words and the separate odd and even information words are then delayed by a preselected time period. The odd and even information words are coded according to an error correction code, and the delayed and coded odd and even information words are then combined in the transmission block. The odd and even sequences respectively comprise N words, and each coding operation is effected by generating an error correction word from the N information words.

Description

PCM signal transmission method

1 is a block diagram showing the main parts of a recording system of one embodiment applied to the PCM recorder of the present invention.

2 is a block diagram showing the main parts of a playback apparatus of an embodiment of the present invention.

3 is a timeline diagram illustrating signal processing near an edit point in one embodiment of the present invention.

4 and 5 are block diagrams of a specific example of a glock device using a crossword code.

6 and 7 are diagrams used for explaining an example of this recorder.

8 and 9 are block diagrams of a specific example of a playback apparatus using a crossword code.

10 is a diagram used for explaining an example of this playback apparatus.

11 is a timeline diagram used for explaining signal processing when reproducing a splice-edited magnetic tape recorded from an example of a recording apparatus.

12 and 13 are block diagrams of another embodiment of the recording apparatus each using a crossword code.

14 is a block diagram of an example of a recording apparatus using an adjacent encoder.

15 is a block diagram of an example of a playback apparatus using an adjacent encoder.

The present invention relates to a method of transmitting a PCM signal, and more particularly, to a method of transmitting a PCM signal in which a plurality of information words are recorded for transmission and then encoded with an error correction code having a significant error correction capability so that the original information word is reproduced. It is about. When transmitting or receiving a PCM signal, noise and signal loss occur at the transmitting and receiving device or the transmitting device. Therefore, encoding has been attempted to correct an error occurring at the receiving side, and the PCM signal is recorded on a recording medium (magnetic tape or disk). The same applies to playback. Correction of Error There is a limitation in the correction capability of coding, and error correction is necessary when error correction is impossible. For example, the PCM demodulated audio signal needs to be corrected so that an error correction cannot be caused by hearing loss.

SUMMARY OF THE INVENTION An object of the present invention is to provide a PCM signal transmission method adapted to a PCM recorder of a fixed head type in realizing a PCM signal transmission method having excellent error correction capability.

The fixed head type PCM recorder differs from the rotary head type in that the PCM signal is recorded along the length of the magnetic tape. Therefore, the splice for electronically splicing the signal onto the magnetic tape with respect to the vertical direction of the magnetic tape is connected. It has the advantage of being easy to edit. When the magnetic tape on which such splice editing has been performed is reproduced, a lot of errors occur near the edit point, and this error cannot be corrected. In addition, such correctable coding is performed by interleaver array recording in word units, and the data is returned to its original position by the playback side (deinterleave). In the part of the data forming the correction block of the same error is lost, the error correction is impossible.

It has been proposed as a method of calibration in fixed-head PCM recorders, which distributes the signal over multiple tracks and records multiple signals, making it possible to reduce the probability that multiple pieces of data constituting a common error correction block will be wrong at the same time. Use a lot of things. One example of this is to distribute and record a PCM signal and error correction (or detection) code of a predetermined length constituting a block of a cross-correction code in parallel to multiple tracks, and to the data of multiple Weeds in the same width direction. To correct the error. According to this error correction method, in the excitation section near the edit point, not only correction but also error correction are impossible. In this case, a process of discarding a plurality of error sections and accessing the normal signals before and after is performed by the memory. As a result, there is a drawback of losing information near the edit point. In addition, the tape speed should be slightly faster due to the need to compensate for the reduction of data in the memory by discarding multiple error intervals. Thus, changing the tape speed from the normal speed also depends on the frequency of the clock that processes the playback signal. There is a need to change, and accordingly, the playback data has to be synchronized to a predetermined clock frequency, resulting in a complicated timing relationship of the playback device, and moreover, a difficult time to perform time management using a time code or the like. .

Another method of error correction in a fixed-head PCM recorder is to divide and record the main data and the sub word, which are the same data, in a plurality of parallel tracks. Rise, there is a flaw that complicates his hardware. Moreover, a method of shifting the recording positions from each other when recording the main word and the sub word in different tracks or recording them by interleaving the main data and the sub data in the same track has also been proposed. In error correction, when the main word is detected to be wrong, the main word is replaced with a subword. According to such an error correction method, since the corresponding words of the main word and the subword are recorded in different positions, the main word and the subword have the same timing so as to delay one of the reproduced main word sequence and the subword sequence to match. In this case, the error in the vicinity of the edit point is avoided by using the position of a plurality of error sections near the edit point in the main word sequence and the subword sequence cloud. However, there is a drawback that the error correction capability is considerably degraded in different error clouds for each of the main word and the sub word, and the data of different magnetic tapes are mixed because there is a main word sequence and a sub word series relatively close to the edit point. Despite the possibility of error in oil, the drawback is that there is no ability to correct or correct errors in oil.

It is therefore an object of the present invention to provide a PCM recorder which does not need to change the speed of the magnetic tape near this edit point without losing information near the edit point. In addition, the present invention provides a signal PCM transmission method that does not cause an increase in redundancy hardware, unlike a dual recording apparatus. Furthermore, it is also a main object of the present invention to provide a PCM signal recording method having excellent error correction capability and error correction capability near the edit point. Further, according to the present invention, it is not necessary to record a plurality of tracks in parallel, and thus a PCM recorder capable of recording a plurality of channels of PCM signals with a magnetic tape of a limited width can be realized. It can provide the most suitable device for the business purpose.

Next, an embodiment in which the present invention is applied to a fixed head type PCM recorder will be described with reference to the drawings. 1 and 2 separately show the main parts of the recording apparatus and the reproduction apparatus to which the present invention is applied. In FIG. 1, a series of PCM signals of one channel in which one sample of an audio signal is converted into one word, for example, is supplied to the input terminal indicated by (1). Are divided into series. Of these multiple sequences, any 2N words that are in sync have occupied part of the input PCM signal sequence, and are even-numbered words and N-numbered words by the even-numbered distribution circuit (3). Is distributed. The even word sequence of the N PCM signals from the even word is supplied to the error correction encoder 4A, and the odd word sequence of the N PCM signals from the odd word is the delay circuit K of delay amount K. Is supplied to the error correction encoder 4B. As the delay circuit 5, a shift register (placement shifter) or RAM (random address memory) provided for each PCM signal sequence can be used.

The error correction encoder 4A shows (N + M) even data series consisting of N even word sequences and even error word sequences of M error correction codes. Error-correcting codes Odd-word series (N + M) odd-numbered data series appear. 2 (N + M) words included in 2 (N + M) data series of even and odd data series are supplied to the error detection encoder 6 to generate an error detection code therefrom. do. The above 2 (N + M) words and one error detection code are composed of one transmission block and converted into serialized codes for every transmission block, and are supplied as registered signals to one or more transmission blocks. As a result, modulation of 3PM, MFM, and the like is performed and recorded by the fixed head as one track on the magnetic tape.

The reproduction signal from the magnetic tape is parallelized by 2 (N + M) words and one error detection code for each transmission block, and is supplied to the error detection decor 7 as shown in FIG. The error detection decoder 7 judges that when any part of one transport block is wrong, all the words contained in that transport block are wrong. At the same time, the (N + M) even data series is supplied to the error correction decoder 8A, and the (N + M) absorption data series is supplied to the error correction decoder 8A. The determination result (pointer) of the error detection decoder 7 is given for each word and is given to the error correction decoders 8A and 8B, and the correction is performed using the error correction code. The pointer to the word on which the error correction has been performed is erased. N even-data fragments obtained from the error correction decoder 8A are delayed by the delay circuit 9 of the delay amount K. Therefore, the time delay between the even data series and the odd data series added by the recording apparatus is eliminated. As a result, each of the even data series and the odd data series is included, and 2N words of the same timing are included in some sections of the PCM signal in the original order and supplied to the even odd synthesis circuit 10, and the even word series And odd word sequences are alternately present, and are supplied to the error correction circuit 11. The error correction circuit 11 cannot correct an error in the error correction decoders 8A and 8B, thus interleaving a word in which the pointer is not erased to another word adjacent thereto. (interleave: interpolation). As an analog signal such as an audio signal, the level difference between one sample and an adjacent sample is small in most cases, so that interleaving with adjacent circles allows good correction. As interleaving, a cross using an average value of two adjacent words or a cross using a plurality of adjacent words can be interleaved. When the output data from the error correction circuit 11 is supplied to the serialization circuit 12, the output terminal 13 can obtain the PCM signal sequence of one channel in the original arrangement order, and the audio signal is obtained by the PCM demodulator. You can get

In practice, the same processing is performed as for each channel of the PCM signal sequence of multiple channels as well as one channel, and is recorded on the magnetic tape as multiple tracks. For the sake of clarity, assuming that the PCM signal series of one channel splice edits two magnetic tapes based on one track, the magnetic tape 14A at the splice point Ps as shown in FIG. 14B is cut | disconnected, and magnetic tape 14B (indicated by a diagonal area) is connected behind magnetic tape 14A at the splice point Ps. In this way, when the splice edited magnetic tape is reproduced, the even data series and the odd data series supplied to the calibration decoders 8A and 8B in FIG. 2 are shown in FIG. That is, the even data series Ea and the odd data series Oa reproduced from the magnetic tape 14a appear in front of the splice point Ps and the corresponding timing edit point Ts. The even data series Ea and Eb are subjected to error correction processing in the error correction decoder 8A, and the odd data series Oa and Ob are error correction processing in the error correction decoder 8B.

If each data series with such error correction is indicated by adding [], the even data series [Ea] and [Eb] are delayed by K by the delay circuit 9, so that the data is edited in odd data series as shown in FIG. The edit point TsK, which is delayed by K with respect to the point Ts, appears in the even data series. In the interval between the Ts and TsK, the even data series [Ea] of the magnetic tape 14a and the odd data series [Ob] of the magnetic tape 14b appear. ] Will exist. Because it is near the edit point, the section has too much error to be corrected, and the section of (Ts ± α) is also an uncorrectable section. The length of these two calibration gaps (± α) is determined according to the calibration capability and error degree having an error calibration code.

When using interleaving as a form of error correction code, a portion of the data forming the same error correction block is lost as a result of deinterleaving, so that an error correction section having a length related to the delay amount of the interleaving occurs. The error interval may be determined by taking into consideration the length of the error correction impossible section due to such an arc shape and the physical multiple error sections by splice editing. The process of restoring the playback data for each of the magnetic tapes 14A and 14B from the playback data in the time relationship shown in Fig. 3C will be described. As for the magnetic tape 14A, as shown in FIG. 3, an even data series [a] with error correction exists until the timing of (Tsk-α). In addition, an odd data series [Oa] exists with error correction up to the timing of (Ts-α). Therefore, the same reproduction can be performed by both [Ea] and [Oa] until the timing of (Ts-α). . In the following section of Ts-α, correction is impossible, and in the section of Ts- (Tsk-α), no odd data series of the magnetic tape 14A exists. Therefore, in the interval between (Ts-α) and ((Tsk-α), an odd data series Oa '(dash represents interpolation data) is formed by interpolation or interleaving using an even data series [Ea]. .

As for the magnetic tape 14B, as shown in FIG. 3E, there is an odd data series [Ob] which is delayed from (Ts + α) and corrected for error. On the other hand, the even data series does not exist in the interval between (Ts + α) and Tsk, and correction is impossible in the interval between Tsk and (Tsk + α). In the intervals (Ts + α) to (Tek + α), even data series Eb 'is formed by interpolation using odd data series [Ob]. In the section after (Tsk + α), even-numbered data series [Eb] and odd-numbered data series [Ob] exist together, so that the same reproduction can be performed as usual.

As is apparent from FIG. 3 D and E, the regenerated PMC data of the magnetic tapes 14A and 14B coexist in the sections of Tsk + α) to (Tsk-α). Although not shown in FIG. 2, the reproduction PCM data (FIG. 3D) on the magnetic tape 14A and the reproduction PCM data (FIG. 3E) on the magnetic tape 14B are shown from the error correction circuit 11. These reproduction PCM data are made up of individual serial data by the serialization circuit 12. Therefore, the cross-fading circuit (cross-fading circnit) is installed behind the rectifying circuit 12, and cross-fading is performed in the crossfading section, which is an arbitrary length from (Ts + α) to (Tsk-α). The discontinuity of the signal can be alleviated. That is, the level of the reproduction data of the magnetic tape 14A is gradually decreased (feather out) in the cross fading section so as to gradually increase (feather in) the reproduction data of the magnetic tape 14B. This cross fading process can be performed by an integer generator that generates an integer that changes with time and a multiplier that multiplies the PCM data with this integer. Since the interpolation data Oa 'and Eb' are approximated with the original data, or the distance between these existing intervals becomes the cross fading interval, even if the error with the original data is large, it does not affect the hearing.

As an example, the PCM signal of one channel is divided into four (i.e., N + 2) PCM word sequences SW ₀ , SW ₁ , SW ₂ , and SW ₃ by the parallelization circuit ₂ , and error correction encoding 4A (4B). Fig. 4 shows a configuration in which two parity word sequences are formed (M = 2). If n is a transmission block number and corresponding integer, then series SW ₀ is represented by a word of W _4n , series SW ₁ is represented by a word of W _n +1, and series SW ₂ is represented by a word of W _4n +2. The series SW ₃ is represented by the words W _4n +3. The even word sequences SW ₀ and SW ₂ in the even odd distribution circuit 3 are applied to the error correction encoder 4A so that the odd word sequences SW ₁ and SW ₃ are delayed circuits 5A with a K (block time) delay amount. And (52) to the error correction encoder 4B. Of these four word sequences, each word of W _4n , W _4n +1, W _4n +2, and W _4n +3 is synchronized with each other, and all of the error correction encoders 4A and 4B are configured identically.

Adjacent two words from each of two even word sequences (odd word series) W _4n , W _4n +2 (W _4n +1, W _4n +3) Exclusive OR gate (15) Is supplied to generate the first parity word P _4n (P _4n +1). These three words constitute a first parry block. In word line SW ₀ (SW ₁₎ is directed into the same output, a word line SW ₁₂ (SW ₁₃₎ and the other via a delay circuit 161, 171 is directed to the output as the word line SW ₁₂ (SW ₁₃₎ The patch word sequence at the exclusive OR gate 15 is directed to the output as word sequence SP ₁₀ (SP ₁₁ ) via delay circuits 162 and 172. Separately between the word sequence SW (SW ₁ ) and the delay circuit (16 ₁ ) (17 ₁ ) and the delay circuit (16 ₂ ) (17 ₂ ), which are not delayed by the error correction encoder (4A) (4B). as the take-3 word is supplied to the exclusive OR gate 18 is formed from the parity words of the second, the parity word sequence is via a delay circuit (17 ₃₎ generating as an output a word sequence SQ ₁₀ (SQ ₁₁₎ do. The three words supplied to the exclusive OR gate 18 and the parity words of the outputs constitute a second parity block.

Delay circuits 161 ₁ and 162 ₂ are for interleaving two PCM word sequences and a first parity word sequence. When the unit delay amount is d (block time), the delay amounts are respectively d and 2d. Has been selected. The delay circuits 17 ₁ , 17 ₂ , and 17 ₃ are for interleaving three PCM word sequences and the first and second parity word sequences, and the unit delay amount is D (block time). The delay amounts of Dd, 2 (Dd), and 3 (Pd) are set separately. The delay amount K (K> 3D-d) of the above-described delay circuits 5 ₁ and 5 ₂ is selected. In the error correction encoder 4A, the PCM word sequence SW ₀ represented by the word W _4n and the PCM word sequence SW ₁₂ represented by the word W ₄ (nD) +2 and the parity word P ₄ ( The first parity word series SP ₁₀ and the parity word series SQ ₁₀ generated by n-2D) are generated. In the error correction encoder 4B, the PCM word sequence SW ₁ represented by the word W ₄ (nK) +1 is generated. PCM word sequence SW ₁₃ represented by word W ₄ (nDK) +3, word P ₄ (first parity word sequence SP ₁₁ represented by n-2D-K (+1), and word Q ₄ (n-3D A second bareness word sequence SQ ₁₁ is generated, denoted by -R + 3d) +1. Eight words extracted from each of these eight word sequences are sent to the CRC generator 6 from the error detection encoder. This 8 word and CRC code are not shown but the sync signal consists of 1 transmission block. It is converted into a soft cord and recorded by the fixed head as one track on the magnetic tape.

Cyclic redunclancy check (CRC) is a method of error detection by means of a traversal code. The 8-word data in the entire block is represented by a polynomial on the GF (2). The surplus is added as a CRC code and recorded. On the reproduction side, error detection is performed by subtracting the polynomial according to the data and the CRC code into the above generated polynomial. The CRC has typically determined that if any word in the same transmission block is wrong, all other words in that transport block are wrong even though it is right. However, according to the method proposed by the applicant first, it is also possible to search for the wrong one word or two words even under the condition that only one word (or two words) are wrong in the same transport block by using the remainder generated in the CRC on the reproduction side. Do.

Furthermore, in order to delay K for the odd word sequence and its associated parity word sequence, a delay circuit 5 ₁ (5 ₂ ) (5 ₃ ) is provided at the output side of the error correction encoder 4B as shown in FIG. 5. (5 ₄ ) may be provided. Alternatively, the even word sequence may be delayed by R rather than the odd word sequence.

For example, when (K = 55) (D = 16) (d = 2), the delay amounts of the delay circuits 16 ₁ , 16 ₂ , 17 ₁ , 17 ₂ , and 17 ₃ are as follows. It becomes.

d = 2, 2d = 4

D-d = 14 2 (D-d) = 28, 3 (D-d) = 42

In selecting a delay amount based on these values, the output sequence SW ₀ , SW ₁₂ , SP ₁₀ , SQ ₁₀ , SW ₁ , SW ₁₃ , and SP ₁₁ SQ ₁₁ from the error correction encoders 4A and 4B. Each will include a PCM word and a parity word in the same order as in FIG. 6 for block numbers from (n = 1 to n = 55). For example, in case of (n = 0), as shown in FIG. 7, a 4-word PCM word (W ₀ , W-62, W-219, W-281) and a 4-word parity word (P-128, Q) -168, P-347, Q-387) and a signal serialized with the signal registered in the CRC code are recorded on the magnetic tape.

8 shows a configuration of the reproduction apparatus corresponding to the configuration of the recording apparatus of FIG. The reproduction signal from the magnetic tape is transmitted to the path indicated by the broken line by the CRC checker 7 for each transport block, and the result is individually transmitted to the error correction decoders 8A and 8B together with the values. The error correction is performed separately for each of the even Deira series (SW ₀ , SW ₁₂ , SP ₁₀ , SQ ₁₀ ) and the odd data series (SW ₁ , SW ₁₂ , SP ₁₁ , SP ₁₁ ).

The error correction decoders 8A and 8B have delay circuits 19 ₁ , 19 ₂ and 19 _{3 for} deinterleaving the input data series (FIG. 9). The delay amounts of the delay circuits 19 ₁ , 19 ₂ , and 19 ₃ are delayed by the delay circuits 17 ₁ , 17 ₂ , and 17 ₃ in the error correction encoders 4A and 4B. The timing of another word sequence is matched to the second delayed parity word sequence SQ ₁₀ (SQ ₁₁ ). That is, the delay amount of the delay circuit 19 ₁ to which the PCM word sequence SW ₀ (SW ₁ ) is supplied is 3 (D-1), and the delay amount of the delay circuit 19 ₂ to which the PCM word sequence SW ₁₂ (SW ₁₃ ) is supplied. and the delay amount to 2 (Pd), is presented to a first series of parity words ₁₀ SP delay amount (Dd) of the delay circuit (19 ₃₎ (SP ₁₁₎ is applied. After this interleaving process, error correction by the second parity word is performed in the Q deckwoor 20. For example, four words (W ₄ (n-3D +) included in the even data series SW ₁₀ , SW ₁₀₃ , SP ₁₀₀ , and SQ ₁₀ supplied to the Q decoder 20 to constitute a second parity block. 3d), W ₄ (n-3D + 2d) +2, P ₄ (n-3D + d), Q ₄ (n-3D + 3d) If only one word is wrong, the above four words (nod.b ), The error word can be corrected by using the syndrome obtained by the addition method according to the addition method, and the pointer to the word is erased for the error correction.Odd data series (SW ₁₁ , SW ₁₀₃ , SP ₁₀₁ , SQ ₁₁₎ The same is true for).

The two word sequences and the first parity word series that have been subjected to the calibration process by the Q decoder 20 are deinterleaved by the delay circuits 21 ₁ 21 ₂ and then supplied to the P decoder 22. The delay amounts of the delay circuits 21 ₁ and 21 ₂ are respectively reduced to (2d) and (d) to eliminate the delay caused by the delay circuits 16 ₁ and 16 ₂ in the error correction encoders 4A and 4B). The timing of the first parity word sequence and the PCM word sequence is matched.

In the decoder 22, three words (W ₄ (n-3D + d) W ₄ (n-3D + d) +2 and P ₄ (n-3D) included in an even data series to form a first parity block When only one word is wrong in + d), the wrong word can be corrected using the synthism obtained from the above three words, and the pointer to the corrected word is erased.

The even word sequence obtained from the error correction coder 8A is passed through the delay circuit 9 ₁ (9 ₂ ) of K (block time) to match the odd word sequence and timing so that the word sequence is an even odd synthesis circuit. Supplied to (10). The error word not corrected by the error correction decoders 8A and 8B is accompanied by a pointer without being erased, thereby correcting an error. For example, any even number of words W _4n +2 when it is invalid and that before and after the correction to andoeeot W _4n +2 is eve interleaving as the average of the odd word W _4n +1 and +3 of the two words of W _4n.

Further, as shown in FIG. 9, at the input side of the error correction decoder 8A, a delay circuit 9 ₁ (9 ₂ ) (9 ₃ ) (9 ₄ ) which delays each of the four even data sequences by K. FIG. It is clear that we can set

In the configuration of FIG. 8, when (K = 55) (D = 16) (d = 2) as described above, an error correction is supplied to the Q decoder 20 in the decoder 8A. SW ₁₀ , SW ₁₀₂ , SP ₁₀₀ , SQ ₁₀ and the data series SW ₁₁ , SW ₁₀₃ , SP ₁₀₁ , SQ ₁₁ supplied to the Q decoder 20 in the error correction decoder 8B are (n). = 42) to (n = 51), the words in the same order as in FIG. 10 are included. In FIG. 10, four words are combined as a second parity block with a horizontal dotted line, and three words are combined with a diagonal line as a first parity block. As can be understood from the 10th degree, each word of the PCM signal includes two different parity blocks, so even if the correction is impossible by one parity block, it may sometimes be corrected by the other parity block. The case can be made small. Originally, a plurality of words constituting the first and second parity blocks are recorded in a distributed position on the magnetic tape by interleaving. Thus, when deinterleaving, two or more words in the same parity block are wrong and cannot be corrected. The odds are low. Therefore, the error correction code described above is called a cloth interleaved code due to its excellent error correction capability.

The configuration of the error correction decoders 8A and 8B of the cloth interleaved code shown in Figs. 8 and 9 is the most basic, and various modifications are possible to further increase the error correction capability. One of them is a method of performing the error correction by the first and second parity words twice. That is, a new P decoder is installed behind the P decoder 22, the output of the Q decoder 22 and the second parity word sequence in the Q decoder 20 are delayed and supplied to the new decoder. The output from the orderer and the first parity word sequence are delayed and supplied to the new decoder. The other method involves error, so that the syndrome formed by the remaining Q decoder or P decoder generated in the CRC checker 7 can be compared with the auxiliary syndrome formed by dividing by the polynomial. There is an error correction decoder (referred to as a crossword decoder) that processes the error word. Moreover, error correction is possible by combining these methods.

How signal processing is performed by a playback device (FIG. 8 or 9) when splicing two magnetic tapes on which a PCM signal is recorded using the above-described recording device (FIG. 4 or 5). Will be described with reference to FIG. It is basically the same as in the case of FIG. 3, except that encoding by interleaving is performed. At the edit point Ts, the result of the deinterleaving in the error correction decoders 8A and 8B, each of the parity word sequences SQ ₁₀ and SQ ₁₁ changes from being reproduced on the first magnetic tape to that on the subsequent magnetic tape. . As shown in Fig. 11A, when the reproduction data series on the subsequent magnetic tape are shown by diagonal lines, playback from all the magnetic tapes is performed in the section (Ts to Ts + d) of the longest delay amount (3D-d) from the edit point Ts. Since data is mixed, the parity block is not completed by the reproduction data from the same magnetic tape, so error correction cannot be performed.

In addition, the PCM even-word sequence processed by the error correction decoder 8A and delayed by K is shown in FIG. 11 so that the error correction can be performed from the delayed edit point Tsk to (3D-d) delay timing (Tsk + α). It becomes period that cannot be. As for the PCM odd word sequence from the error correction decoder 8B, the section of (Ts to Ts + α) becomes an uncorrectable period as shown in FIG. Thus, as for the first magnetic tape, as shown in FIG. 11B, even word sequences [Ea] and odd word sequences [Oa] are obtained by different error corrections up to the edit point Ts (Ts to Tsk). The interpolation odd word series Oa, by the word series [Ea], is formed. As for the later magnetic tape, as shown in FIG. 11E, even word sequences Eb 'are obtained by interpolation by odd word series [Qb] in the section of (Ts + α to Tsk + α), and (Tsk + α). In later sections, even-word sequences [Eb] and odd-word sequences [Ob] are obtained by further error correction. Cross fading processing is performed in a section (Ts + alpha-Tsk) where the reproduction PCM words of these two magnetic tapes overlap.

In addition, the above description focuses on the section in which the error of the cross interleaved code is not corrected, and there exists a section having more error before and after the edit point. Multiple error intervals after the edit point Ts (Tsk) are included in the interval of (3D-d) which is determined by the cloth interleaved code itself. In addition, there are a plurality of error intervals before the edit point Ts (Tsk), and it is necessary to form the odd denita series Oa by interpolation of the even data series [Ea] before the edit point Ts. Due to a large number of error intervals, the odd data series Oa 'cannot be delayed to the delayed edit point Tsk.

12 and 13 show another modified configuration of the recording apparatus. The recording apparatus shown in FIG. 12 divides the PCM signal sequence of one channel into 12 word sequences (SW ₀ to SW ₁₁ ), supplies six even word sequences to the error correction encoder 4A, and simultaneously delays six odd word sequences. The circuits 5 ₁ to 5 ₆ are delayed by K (block time) to be supplied to the error correction encoder 4B, and the first parity word sequence and the second parity for each word sequence are provided. It is a configuration forming a word series. That is, when (N = 6) (M = 2). The error correction encoders 4A and 4B are an exclusive part OR gate 15 which forms the first parity word from six words of the input word sequence and a delay circuit 16i (i) of block time. = 1,2,3,4,5, and the exclusive OR gate 18 forming the second parity word and the delay circuit 17j (j = 1,2) of j (Dd) (block time) , 3, 4, 5, 6, and 7, so that each of the error correction encoders 4A and 4B generates a twelfth PCM word sequence and four parity series. The 16 words extracted from the 16 word sequences are formed by the CRC generator 6 into CRC codes. Therefore, one transmission block recorded on the magnetic tape is composed of twelve PCM words, four parity words, a CRC code, and a synchronization signal.

When such a recording device as shown in FIG. 12 is used, the same effect as described above is obtained. In addition, redundancy can be reduced by increasing the number of PCM words included in one transport block, which becomes more practical.

In the above recording device, one PCM signal series is divided into N even and odd under sequences, respectively, and each of these N word series is supplied to separate error correction encoders, and the error correction encoders are even and odd words on the input side. It doesn't need to be divided into series. As shown in FIG. 13, the PCM word sequence SW ₁ including the even word W _4n and the odd word SW ₁ +1 alternately includes the PCM word sequence SW ₂₃ including the even word W _4n +2 and the odd W _4n +3. The first parity in which the even parity word P _4n and the odd parity word P _4n +1 alternately exist by the exclusive OR gate 15 so that even and odd words in the PCM have the same timing. Form a word series. The exclusive OR gate 18 forms a second parity word sequence in which even parity word Q _4n and odd verity word Q _4n +1 are alternately present. Even word and odd word and odd number by the switching circuit 23 provided on the output side of the error correction encoder 4 while performing interleaving by the delay circuits 17 ₁ , 17 ₂ , and 17 ₃ . By separating the words, the separated odd word sequences are delayed by K by the delay circuits 5 ₁ to 5 ₄ .

W_4n+2P₄(n-k)+1=W₄(n-k)+1

W₄(n-k)+3의 제 1 의 패리티 워드가 형성되는 것은 상술한 바와 똑같다. 또 이들의 PCM 워드로 부 형성되는 제 2 의 패리티워드는 Q_4n=T W_4n

+2Q₄(n-k)+1=T²W₄(n-k

TW₄(n-k)+3로 된다. 여기서 T느 1 워드의 비트길이 실예로 16비트로 했을 때에, (16×16)의 생성 다항식(매트릭스)를 나타낸다. 즉,Furthermore, other error correction codes other than the cloth interleaved codes described above may be used. As an example, one configuration of a recording apparatus and a reproduction apparatus when using a matrix code, which is a kind of b-adjacent code, is shown separately in FIG. 14 and FIG. That is, each of the encoders 4A and 4B in the error correction in the recording device is provided with an encoder 24 adjacent thereto, and the first and second parity are provided by the adjacent encoder 24. Word is formed. P _4n = W _4n from each of the PCM words (W _4n , W _4n +2) (W ₄ (nk) +1, W ₄ (n = k) +3)

W _4n + 2P ₄ (nk) + 1 = W ₄ (nk) +1

The formation of the first parity word of W ₄ (nk) + 3 is the same as described above. The second parity word formed by these PCM words is Q _4n = TW _4n.

+ 2Q ₄ (nk) + 1 = T ² W ₄ (nk

TW ₄ (nk) +3. Here, when the bit length of one word is 16 bits, the generation polynomial (matrix) of (16x16) is shown. In other words,

로 나타내지는 GF(2)상의 다항식이며, I₁₅는 15차의 단위 매트릭스이다.to be. The polynomial G (X) is

It is a polynomial on GF (2), and I ₁₅ is a 15th order unit matrix.

The PCM word sequence and the parity word sequence are interleaved by the delay circuits 17 ₁ , 17 ₂ , and 17 _{3 in the} error correction encoders 4A and 4B. The delay amounts of the replay circuits 17 ₁ , 17 ₂ , and 17 ₃ are selected to be D, 2D, 3D, respectively. In addition, a CRC code for error detection formed by the CRC generator 6 is added to each transport block.

The reproducing apparatus detects an error by the CRC checker 7 and supplies a word sequence to the error correction decoders 8A and 8B together with the splitter. The error correction decoders 8A and 8B are provided with delay circuits 19 ₁ , 19 ₂ and 19 ₃ (having 3D, 2D and D delays respectively) for de-interleaving. The interleaved PCM word and the first and second fate words are supplied to the adjacent decoder 25 to perform error correction. In the adjacent encoder 25, the error of one word among those included in the same first parity block can be completely corrected. The detailed description is omitted here, but since the specific error word can be designated by the CRC as well as the error of two words included in the same second parity block, it can not be corrected by the adjacent encoder 25. have.

As described above, it is a kind of adjacent code of the matrix code that multiplies the code symbol by the generation matrix. Adjacent codes are symbols of GF (2b) (2D Golois fielcl) and can be used to correct the error of the bit group, and are generalized Hamming code and Reed Solomon code in addition to matrix codes. And so on. Adjacent codes have the advantage of high hardware complexity but error correction capability. As described above, the PCM even-wedding sequence from the error correction decoders 8A and 8B passes through the delay circuits 91 and 92 of the (block time).

When splice editing a magnetic tape on which a PCM signal has been recorded using the recording apparatus shown in FIG. 14 described above, data recorded on the front and rear magnetic tapes can be reproduced in the same procedure as in FIG. Error correction cannot be made to the cloud after the edit point Ts (Tsk) (3D). Furthermore, since the ability to correct adjacent codes is high enough to correct errors of two words in the same parity block as described above, even errors contained in the cloud of (3D) after the edit point Ts (Tsk) can be corrected. Often there are.

As can be understood from the above description, the present invention has the following effects.

First, error correction coding can be performed on each of the even word sequence and the odd word sequence, so that when the word contained in either the even word sequence or the odd word sequence becomes uncorrectable, it is adjacent to the error word and on the other side. Good interpolation can be performed as the correct word included, resulting in an error correction capability. In addition, there is no drawback that the redundancy is large or the hardware is complicated, such as a device for double recording in which two words of the same data are recorded at different positions on the magnetic tape.

Secondly, when the present invention is applied to the fixed-head type PCM recorder, not all track recordings are required for error correction coding, so that one channel can be recorded as one track, so that a magnetic tape of a predetermined width can be recorded. It can record PCM signals of a very large number of channels. Therefore, a PCM recorder useful for business purposes can be realized.

Third, when the present invention of the fixed-head type PCM recorder is applied, a good process can be performed for a large number of error sections near the edit point of splice editing. That is, since the data of a large number of error intervals near the edit point are not discarded, there is an advantage that there is no loss of information and the complexity of timing control that changes the tape speed at a normal speed is not caused. In addition, by using this recording method, a large number of error intervals occur at different timings of the main word series and the western word series. The error correction capability is lost and the error correction capability is also lost. However, according to the present invention, the error correction for each word sequence of each magnetic tape can be performed even in a section where the data of the changed magnetic tape is mixed as an even word sequence and an odd word sequence. Therefore, the error can be corrected by using the negated result. Furthermore, the present invention can be applied to an optical PCM disk that reads a pint (concave) on a rotating disk by a laser beam or a PCM disk that reads a change in irregularities on a rotating disk as a change in capacitance or an output of a piezoelectric element. . In addition, the error detection code may use a pyritic code other than the CRC code, a full adder code, or the like.

Claims (1)

In a method of transmitting a PCM signal that becomes a digital information word, the information is divided into an odd information word sequence and an even information word sequence, and then the separated even odd information words are time-shifted with respect to each other by a predetermined amount of difference. The odd-correction code is encoded by the error correction code, and the even-numbered information word is encoded differently from the odd information by the error correction code, and the even-numbered information word encoded by the time shift is combined into the transport block. PCM signal transmission method comprising the step of transmitting the transport block.

Images