KR100244768B1 - Asynchronous partial responsive data detector of a digital magnetic recording/reproducing system - Google Patents

Abstract

The present invention relates to an asynchronous partial response data detection apparatus for estimating code data recorded on a magnetic medium from digital data sampled from coded data reproduced from a magnetic medium in a digital VCR. Adaptive equalization means for outputting as an equalization signal having partial response channel characteristics in accordance with the coefficient value; Phase control means for synchronizing a sampling time of the digital data and a code time difference of the code data by detecting and compensating for a phase error using the equalization signal; The level of the ideal equalization signal of the first and second states, respectively selected according to the first and second selection signals, is determined as the sign data for the equalized signal, wherein the selected sign data is the first and second previous signals. Each state starts from the two possible states of and includes the highest order detector, which is data that survives in such a way that each state transitions to one of the two states.

Description

Asynchronous Partial Response Channel Data Detection System for Digital Magnetic Recording / Playback System

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital magnetic recording / reproducing system (hereinafter referred to as " digital VCR "), and more particularly to an asynchronous data detection apparatus in a digital VCR of a partial response channel system.

A digital VCR for recording a digital signal on a magnetic tape, comprising a recording mode for magnetizing a magnetic medium such as a magnetic tape through a wound magnetic head and a rotary transformer, and amplifying a magnetic flux change of the magnetic tape and restoring or decoding them into digital data. There is a playback mode.

Digital VCR cannot transmit direct current component, which is the basic characteristic of magnetic channel, by using magnetic medium as transmission channel, and inter-symbol interference (ISI) due to high frequency attenuation and high density recording Interfere with the transmission. In addition, the tape may be affected by the feeding of the tape and the rotation control of the head.

In order to reduce the error rate in the magnetic recording / reproducing path, a conventional digital VCR uses a partial response channel IV (partial response class IV). Such partial response channel is processed so that the transmission channel has characteristics similar to the medium in consideration of the characteristics of the magnetic medium, so that the magnetic channel response is approximated to the ideal partial response IV channel, thereby reducing the error data when the recording data is restored. In this way, in order to amplify the magnetic flux variation of the magnetic tape to completely recover the digitally recorded signal, the reproduction signal must be made into a signal having partial response channel characteristics.

Digital data detection systems for detecting two values of signals reproduced in a digital VCR using a conventional partial response channel mainly use an adaptive decision feedback equalizer (DFE). The DFE performs a function of finding an optimal weight for the distorted reproduction signal through the channel and using the same to restore the original desired signal.

In a digital data detection system having a decision feedback equalizer, a reproduction signal from a recording medium is passed through an adaptive feedforward equalizer (FFE), and then through an adaptive decision feedback equalizer. The sum of the equalizers creates a record / play channel response that meets the ideal partial response IV channel response characteristics. Therefore, the output signal from the decision feedback equalizer means the data signal generated by the reproduction signal passing through the ideal partial response channel.

If one models the digital VCR system, a partial response channel, the reproduction restoration of the digital signal is ternary values {y _n by determining the case in preparation for the threshold value (threshold) preset the signal amplitude of the analog reproduction data from any identified point } Determined as one of (+2, 0, -2), and the processed three-value value is decoded into data {b _n } (0, 1) according to a binary decoding decision rule as shown in the following equation for each data bit period. Is processed.

Recently, the digital data detection apparatus for further improving the above-described digital data detection apparatus uses a maximum likelyhood sequence detector in conjunction with the output of the adaptive decision feedback equalizer to reduce the error rate. The highest order detector is an apparatus for estimating the code sequence that is closest to an arbitrary received code sequence, and uses a Viterbi algorithm which effectively searches for the above-described close code sequence using trellis. The Viterbi detector does not estimate the received data for each data in the above-described prior art data detection apparatus but estimates using some previously received data.

The Viterbi detector maintains an order of data determination, called a survival sequence, and corresponding error metric, for each possible state assumed by the grid at any data time. For example, if there are two states in the grid, we maintain two survival orders and the corresponding two error metrics.

At each data time, the equalized received signal values are evaluated to update the error metric for the state of the grid, resulting in an updated survival order. The updated survival order with the minimum error metric is considered to be the recorded data signal order. The oldest data in the survival order is produced at the output of the Viterbi detector, repeating this method for the next data. Thus, the Viterbi detector can provide a very good decoding efficiency with a lower error rate than the conventional decoding scheme using threshold values.

However, in the conventional data detection apparatus using a Viterbi detector, when the magnetic recording / reproducing system is modeled as a partial response channel system, it is possible to decode the equalized signal passing through the partial response channel directly using the Viterbi algorithm. However, at this time, since the number of states of the grating is four, the calculation included in the algorithm becomes complicated, and accordingly, there is a problem that the circuit configuration of the Viterbi detector is complicated.

Therefore, an object of the present invention is to provide a digital data decoding apparatus of a digital VCR.

Another object of the present invention is to provide a digital data detection apparatus having an improved Viterbi detector in a digital VCR.

It is still another object of the present invention to provide a digital data detection apparatus capable of performing digital data detection in a digital VCR in an asynchronous mode.

According to the present invention for achieving the above object, an asynchronous partial response channel data detection apparatus for estimating code data recorded in the transmission channel from digital data sampled from code data reproduced from the transmission channel is adapted to: adapt the digital data. Adaptive equalization means for outputting as an equalization signal having a partial response channel characteristic according to a typical filter coefficient value; Phase control means for synchronizing the sampling time of the digital data and the code time difference of the code data by detecting and compensating for a phase error using the signal output from the equalizing means; The level of the ideal equalization signal of the first and second states, respectively selected according to the first and second selection signals, is determined as the sign data for the equalized signal, wherein the selected sign data is the first and second previous signals. Each state starts from the two possible states of and includes the highest order detector, which is data that survives in such a way that each state transitions to one of the two states.

1 is a schematic block diagram of an apparatus for detecting data of a digital magnetic recording / reproducing system according to a preferred embodiment of the present invention;

2 is a detailed configuration diagram of a phase detector and a loop filter;

3 is a detailed configuration diagram of an interpolation filter;

4 is a detailed configuration diagram of an adaptive feedback equalizer;

5 is a detailed configuration diagram of the filter coefficient updating unit;

6 is a detailed configuration diagram of the highest order detector;

7 is a detailed configuration diagram of a difference evaluation amount delay unit;

8 is a detailed configuration diagram of a survival order determiner;

9A and 9B show state transition diagrams and trellis of a partial response channel system, FIG.

10A and 10B illustrate the process by which survival paths converge;

11 is an explanatory diagram showing a difference between a sampling time and a data time at an arbitrary time;

12 is a waveform diagram illustrating an ideal response characteristic curve of a finite response filter.

<Description of the code | symbol about the principal part of drawing>

20: interpolation filter 30: feedforward equalizer

40: decision feedback equalizer 55: error signal generator

60: highest order detector 70: phase detector

80: loop filter

The above and other objects and various advantages of the present invention will become more apparent from the preferred embodiments of the present invention described below with reference to the accompanying drawings.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1, a block diagram of an asynchronous partial response channel digital data detection apparatus of a digital VCR according to the present invention is shown. As shown, the asynchronous partial response channel type digital data detection apparatus of the present invention includes a head amplifier 10, an analog / digital converter &quot; 16 &quot;, an interpolation filter 20, and the like. A phase control loop 100, an adaptive decision feedback equalizer 200, a maximum likelihood sequence detector (MLSD) 60, and an MCU 90 which provides the control signals and clocks needed for the data detector. do.

The signal recorded on the magnetic tape in the digital VCR is reproduced by a pair of video heads (not shown) which are alternately switched on the head drum, and amplified by the head amplifier 10. The A / D converter 16 samples the signal amplified by the head amplifier 10 according to a fixed clock and provides a sampled reproduction signal to the phase control loop 100.

The phase control loop 100 has a configuration for correcting a phase difference with actual data in order to synchronize asynchronous timing of random sampling regardless of data timing. The phase control loop 100 includes an interpolation filter 20, a phase detector 70, and a loop filter 80. The interpolation filter 20 controls and restores sampling timing according to the phase difference signal provided from the loop filter 80. Specifically, the difference between the data value sampled at the sampling point by the A / D converter 16 and the data point actually recorded on the tape, for example, the phase difference or the time difference is corrected. The data corrected by the interpolation filter 20 is provided to a feedforward equalizer (FFE) 30.

The phase detector 70 detects a phase change amount as a phase error using the surviving signal from the highest order determiner 60, removes a high frequency component through the loop filter 80, and transfers the high frequency component to the interpolation filter 20. The loop filter 80 increases the loop system performance by keeping the locked state of the phase locked loop 100 stable by removing high frequency components or noise from the output of the phase comparator.

The adaptive decision feedback equalizer 200 includes a feedforward equalizer (FFE) 30, an FFE coefficient updater 35, a decision feedback equalizer (DFE) 40, and a DFE coefficient updater 45. It is composed of

The FFE 30 and the DFE 40 respectively provide a recording / reproducing channel response that satisfies the ideal PR4 channel response of the signal reproduced from the recording medium according to the coefficient values transmitted from the corresponding coefficient updating units 35 and 45, respectively. Perform the create function. Therefore, the signal output from the FFE 30 and the DFE 40 means a data signal equalized by the reproduced signal passing through the ideal partial response channel. The output signals of the FFE 30 and the DFE 40 are added in the adder 50, respectively. The output of the adder 30 is input to the error signal generator 55 and the highest order detector 60. The output of the highest order detector 60 is fed back to the DFE 40 to form a feedback loop. This feedback loop serves to suppress mutual interference between residual data in the recording / reproducing channel. The error signal generator 55 generates an error signal &quot; e &quot; which is the difference between the equalized signal from the adder 50 and the ideal signal detected by the highest order detector 60. This error signal is used to adjust the filter coefficients of the FFE 30 and the DFE 40.

The maximum likelihood determiner 60 uses the highest likelihood decoding algorithm in a manner of estimating the code sequence closest to any received code sequence. The maximum likelihood decoding algorithm includes a Viterbi algorithm that efficiently searches for the above-described close code sequence using trellis. This Viterbi algorithm is effectively used to improve the performance of a data detector with inter-symbol interference.

First, the encoding process using the partial response channel will be described.

＆quot; 은 하기 수학식에 따라 프리코딩(precoding)되어 프리코드된 (precoded) 데이터 열 ＆quot;

＆quot; 로서 생성된다.Binary data sequence &quot;

&Quot; Is a precoded and precoded data string &quot;

&Quot; Is generated as:

In the above equation, 1s represents an exclusive OR operation.

은 하기 수학식에 따라 2극(bipolar) 펄스 열

로 변환된다. 이때, 2극 펄스 열의 데이터는 진폭을 가지고 있다.Precoded Data Column

Is a bipolar pulse train according to the equation

Is converted to. At this time, the data of the two-pole pulse train has an amplitude.

은 하기 수학식에 따른 부호화 규칙에 따라 부호화되어 자기 테이프에 기록되고 그로부터 재생된다.Then, a two-pole pulse train

Is encoded according to an encoding rule according to the following equation, recorded on a magnetic tape, and reproduced therefrom.

는 디지탈 필터인 등화기를 이용하여 재생된 채널 데이터를 나타낸다. 상술한 부호화 과정에 있어서, 프리코드된 데이터 열

은 2진수에 대하여 실행되지만,

을 생성하는 디지탈 필터링은 2극 펄스에 대하여 실행된다는 것이다.In the above formula

Denotes channel data reproduced using an equalizer which is a digital filter. In the above-described encoding process, the precoded data string

Runs on binary numbers,

Digital filtering to generate is performed on the bipolar pulse.

이 ＆quot;0 1 1 0＆quot; 인 경우에 수행되는 부호화 과정을 작성하면 다음과 같다.For example, input symbol (data)

0 &quot; 0 1 1 0 &quot; If the coding process performed in the case of is as follows.

: * 0 1 1 0Binary symbol data

: * 0 1 1 0 Precoded Data Column

: * One One 0 One Precoded Data Column

: One One 0 One One 2-pole pulse data string

: * +1 +1 -One +1 2-pole pulse data string

: +1 +1 -One +1 +1 Equalized Channel Data

: 0 -2 +2 0

은 2극 펄스 데이터 열

와

사이의 관계로부터 구한것이고, 2진 데이터 열

의 값은

와

사이의 관계로부터 구한 것이며, ＆quot;*＆quot;는 돈 케어 상태(don't care)를 나타낸다.Where the precoded data column

Silver 2-pole pulse data string

Wow

From the relationship between the binary data columns

The value of

Wow

The &quot; * &quot; represents a don't care.

가 ±1 인 경우에,

는 세가지의 가능한 값, 즉, +2, 0, -2 의 세가지 레벨(three-level)중의 어느 한 값을 갖는다.Binary in the decoding process

If is ± 1,

Has any of three possible values, namely, three-levels of +2, 0, and -2.

In the encoding process of the partial response channel IV system, the even (or odd) signal code string is only affected by the even (or odd) code. Thus, the partial response channel system may be considered as two independent partial response channel systems with their partial response channel characteristics 1-D. By using this, it is possible to simplify the calculation required to perform the algorithm by dividing the reproduction signal code string into odd and even columns and detecting the data by the VTV decoding method separately. Therefore, the number of states of the grid included in the encoding process of the partial response channel system can be reduced to only two.

라하고, 출력 심볼을

라 하면 이들의 관계는 도 9A에 도시된 상태 전이도와 도 9B에 도시된 격자(trellis)로 표현될 수 있다.Input symbol

And the output symbol

This relationship can be expressed by the state transition diagram shown in FIG. 9A and the trellis shown in FIG. 9B.

9B is a grid for one de-interleaved symbol stream of the partial response channel with P (D) = 1-D. This grating helps to understand the operation of the likelihood order determiner 60. In FIG. 9B, upper vertices 251 and 252 represent state 1 in the grid and lower vertices 261 and 262 represent state 2. In FIG. Left vertices 251 and 261 represent past states in the grid, and right vertices 252 and 262 represent new states. Each branch that progresses from the old state 251, 261 to the new state 252, 262 is shown by an arrow. Associated with each state vertex is an error metric (EM). EM1c is associated with state 1 of vertex 251 and EM2c is associated with state 2 of vertex 261. Associated with each new state vertex is the updated error metric. EM1 is associated with new state 1 of vertex 252 and EM2 is associated with new state 2 of vertex 262.

) 을 받았다면, 가지를 따라 격자상의 다른 상태로 움직이고, 등화기가 2 진 심볼 데이터 0(=

) 를 받았다면, 가지를 따라 같은 상태로 진행한다. 상태의 전이를 일으키는 이상적인 수신된 등화 신호의 값은 입력 신호를 수신했을 때 진행하게 될 격자상의 가지에 대괄호로 표시되어있다. 따라서, 상태 1의 정점 (251)에서 시작하여, 등화된 채널 데이터가 0 이면, 다음의 기호 시간에 상태 1에 남게되고, 정점(252)에서 끝나게된다. 이것은 등화기가 2 진 심볼 데이터 (

)로서 이진수 0을 수신하였음을 의미한다. 등화된 채널 데이터가 값 +2를 가지면, 상태 2로 전이되고, 정점(262)에서 끝난다. 이것은 등화기가 2 진 심볼 데이터 (

)로서 이진수 1을 수신하였음을 의미한다. 마찬가지로, 상태 2의 정점(261)에서 시작하여, 등화된 채널 데이터가 0 이면, 다음의 기호 시간에 상태 2에 남게되고, 정점(262)에서 끝나게된다. 이것은 등화기가 2 진 심볼 데이터 ()로서 이진수 0을 수신하였음을 의미한다. 등화된 채널 데이터가 값 -2를 가지면, 상태 1로 전이되고, 정점(252)에서 끝난다. 이것은 등화기가 2 진 심볼 데이터 (

)로서 1을 수신하였음을 의미한다.From any particular past state, the grid is binary symbol data 1 (=

), Move along the branch to another state on the grid, and the equalizer equals binary symbol data 0 (=

), Proceed in the same state along the branches. The value of the ideal received equalization signal that causes the transition is indicated by square brackets on the branch on the grid that will proceed when the input signal is received. Thus, starting at vertex 251 in state 1, if the equalized channel data is zero, it remains in state 1 at the next symbol time and ends at vertex 252. This means that the equalizer has binary symbol data (

) Means that binary 0 was received. If the equalized channel data has a value of +2, it transitions to state 2 and ends at vertex 262. This means that the equalizer has binary symbol data (

Means that binary 1 has been received. Similarly, starting at vertex 261 in state 2, if equalized channel data is zero, it remains in state 2 at the next symbol time and ends at vertex 262. This means that the equalizer has binary symbol data ( ) Means that binary 0 was received. If the equalized channel data has a value of -2, it transitions to state 1 and ends at vertex 252. This means that the equalizer has binary symbol data (

Means 1 has been received.

)를 입력 신호로서 수신하기는 불가능하다. 따라서, 격자상의 각 가지에서, 가지 메트릭(branch metric) BM 이 계산된다. 각 가지에서, 실제의 수신된 등화된 채널 신호와 그 가지에 해당하는 이상적인 입력 신호간의 크기의 차이가 계산된다. 이 크기를 가지 메트릭 (BM) 이라 지칭하고, 과거에서 누적된 오류 메트릭과 같이 축적된다. 여기서 오류 메트릭은, 각각의 갱신될 오류 메트릭을 만들기 위한 각각의 그 가지에서의 원천 정점(source vertex)에 관련된다. 이 갱신된 오류 메트릭들은 가장 유망한 순서(most likely sequence)를 결정하는데 사용된다.However, real detectors are ideal for binary symbol data (

) As an input signal is not possible. Thus, for each branch on the grid, a branch metric BM is calculated. In each branch, the difference in magnitude between the actual received equalized channel signal and the ideal input signal corresponding to that branch is calculated. This magnitude is referred to as a branch metric (BM) and accumulates like error metrics accumulated in the past. The error metric here relates to the source vertex at each branch to create each error metric to be updated. These updated error metrics are used to determine the most likely sequence.

For the new symbol, there are two ways to reach the vertex 321 representing state 1. Receive a 0 signal at vertex 251 or a -2 signal at vertex 261. The branch metric BM1 is calculated for branches going from vertex 251 to vertex 252 and the branch metric BM3 is calculated for branches traveling from vertex 251 to vertex 252. These branch metrics accumulate with each error metric associated with their source vertex. That is, the branch metric BM1 accumulates like the error metric EM1c accumulated in the past, creating an updated first error metric (EM1 '= BM1 + EM1c) for the branch going from vertex 251 to vertex 252. Serve The branch metric BM3 accumulates with the current error metric EM2c, producing an updated second error metric (EM1 '' = BM3 + EM2c) for the branch going from vertex 261 to vertex 252. Where the updated error metric with the lesser of the first and second error metrics, EM1 'and EM1' ', is assumed to be a more promising branch to reach state 1, and the updated error for vertex 252 It is selected as the metric EM1.

Similarly, the branch metric BM2 is calculated for the branch that progresses from vertex 251 to vertex 262 and BM4 is calculated for the branch that progresses from vertex 261 to vertex 262. Branch metric BM2 accumulates with the current error metric EM1c to produce an updated error metric (EM2 '= BM2 + EM1c) for the branch that progresses from vertex 251 to vertex 262. Branch metric BM4 accumulates with the current error metric EM2c to produce an updated error metric (EM2 &quot;: = BM4 + EM2c) for the branch that progresses from vertex 261 to vertex 262. An updated error metric with a lower value is assumed to be a more promising branch to reach state 2, resulting in an updated error metric EM2 for vertex 262. In other words, vertices with fewer error metrics become the most promising states. This most promising state accumulates and, as shown in FIG. 10, only the surviving path is selected while blocking the unconnected survival path.

Referring now to FIG. 2, a detailed circuit diagram of the phase detector 70 and loop filter 80 of the phase feedback loop is shown.

The phase detector 70 performs a function of detecting a phase error using a signal surviving in the highest order detector 500. The set of three survivals storing a signal surviving the highest order detector 500 is stored. Signal storage registers 610, 620, 630, adder 640, and multiplier 650.

The outputs of the first register 610 and the third register 630 are respectively provided as inputs of the subtractor 640, and the outputs of the second register 620 are provided to the error signal generator 55 which outputs an error signal. . The subtractor 640 subtracts the outputs of the first and third registers 610 and 630 to provide the phase difference signal X _{n + 1} -X _n-1 of the subtracted survival signal to one input of the multiplier 650. . The multiplier 650 outputs the phase error signal Z _k by multiplying the error signal e received as another input signal by the phase difference signal. An operation of the phase detector 70 performing the above-described operation may be expressed by the following equation using a decision directed method.

The phase error signal Z _k is provided to the loop filter 80.

The loop filter 80 maintains the locked state of the phase control loop system by increasing the loop system performance by removing high frequency components or noise from the output of the phase detector 70. The loop filter 80 includes a first multiplier 670, second multiplier 680, a second multiplier 680, which multiplies the filter coefficient β and the phase difference signal Z _k by which to multiply the filter coefficient α and the phase error signal Z _k The first adder 690 adds the output of the output and the output of the delayer 700, and the second adder 710 adds the output of the first adder 690 and the output of the first multiplier 670. The output _1k of the loop filter 80 which performs the above operation may be expressed by the following equation.

The output 1k of the loop filter 80 is provided to the interpolation filter 20 of the next stage.

3, a detailed configuration of the interpolation filter 20 is shown.

The interpolation filter 20 performs a function of restoring timing by asynchronous sampling using a phase error signal. The interpolation filter 20 has a structure of a finite impulse response filter of a digital filter, and the phase error detection unit 800. And a phase error compensator 900.

The phase error detector 800 includes a first adder 810 for adding the phase error signal to W, which is the ratio of the sampling frequency to the actual recording data frequency. Referring to Fig. 11, a waveform diagram illustrating the frequency ratio &quot; W &quot; is shown. In Fig. 10, &quot; Ts &quot; Denotes the level of the reproduced signal sampled at an arbitrary point of &quot; m &quot;, and &quot; Ti &quot; Denotes the level of the signal recorded on the actual magnetic tape at any point of &quot; k &quot;.

In FIG. 3, the output of the first adder 810 is accumulated in the accumulation module 840 through the second adder 820 and the delayer 830. This accumulation is shown in FIG. 11 as the timing increases between the actual recording data point and the sampling data point as time passes. The accumulated value in the accumulation module 840 is directly connected to the first input terminal of the subtractor 860 and is connected to the second input terminal of the subtractor 860 through the integer processing unit 850.

The integer processing unit 850 takes only the integer value from the accumulated value provided from the accumulation module 840 and provides this integer value to the second input terminal of the subtractor 860 to record the data Xn sampled in the subtractor 860 and the actual recorded value. The signal 1k indicating the distance to the data is output to the phase error compensator 900. In addition, the purified water taken by the purified water processor 850 is provided to the comparator 870 and the retarder 875. The comparison unit 870 outputs the delay signal &quot; HOLD &quot; if the difference from the previous integer value via the delay unit 875 is 1 or more. This is when the integer value increases as the values below the decimal point accumulate and occur as a carry.

The phase error compensator 900 includes a lookup table memory 910 and a finite impulse response filter 920. The lookup table 910, which is composed of a memory such as a ROM, stores a response value corresponding to each sampling time as the filter coefficient vector H (t) in the ideal response characteristic curve of the impulse response filter shown in FIG. Each memory area with filter coefficients is addressed by a distance signal 1k. The set of filter coefficients read out from the memory area addressed by the distance signal 1k is provided to the next finite response filter portion 920.

The finite response filter unit 920 multiplies the input signal vector Xn delayed through the tap delay line (not shown) with the filter coefficient vector H (t) read from the lookup table 910, and adds the multiplied values to the interpolated data value. Outputs The output of the finite response filter unit 920 is provided to the FFE 30 (see FIG. 1).

4, a detailed configuration of the adaptive decision noble equalizer 200 is shown.

The FFE 30 is formed in a transversal form, which is a kind of finite impulse response (FIR) structure. That is, a tapped delay line 210 for sequentially delaying the interpolated signal &quot; Xn &quot; (= nth signal) output from the interpolation filter 20 and the delayed signal in each delay line are updated. Adaptive filter coefficient &quot; Cn &quot; provided from section 35; Multiplier block 220 and adder 225 for adding the output of each multiplier in multiplier block 220. The output of adder 225 is provided to the first input of adder 50 (see FIG. 1).

Similarly, the DFE 40 has the same finite response filter structure as the FFE 30. The DFE 40 constructed in the present invention has &quot; 1 &quot;, &quot; 0 &quot;, &quot; -1 &quot; which are data values surviving the highest order detector 60. Because only the values are multiplied, a simple structure can be constructed with fewer gates, which is simpler than a full bit multiplier. The DFE 40 provides a tapped delay line 230 for sequentially delaying the output 65 of the highest order detector 60 and a signal delayed at each delay line from the coefficient updater 45. A multiplier block 240 for multiplying the adaptive filter coefficient &quot; Bn &quot; (= n-th filter coefficient) and an adder 245 for adding the output of each multiplier in the multiplier block 240. The output of adder 245 is provided to a second input of adder 50. The output of the adder 50 is provided to the highest order determiner 60 via line 52 as described with reference to FIG. 1.

5 shows the detailed configuration of the filter coefficient update unit 35, 45. As shown in FIG. The filter coefficient updating units 35 and 45 respectively update the filter coefficients necessary for the FFE 30 and the DFE 40 by using least mean squares (LMS) of the following equation.

The multiplier 202 multiplies the value of the convergence constant 1 by the error signal e. The output of the multiplier 202 is multiplied by the n-1 th input signal vector Xn-1 in the multiplier 204. The output of the multiplier 204 is added with the previous filter coefficient Cn-1 (or Bn-1) provided from the initial filter coefficient updater 214 in the adder 206 to obtain a filter coefficient Cn (or Bn) to be updated. . The filter coefficient Cn (or Bn) updated by the adder 206 is provided to the FFE 30 or the DFE 40 via the hold switch 208.

In the initial state, the initial coefficient updater 214 receives the initial coefficient data according to the initial coefficient load control signal from the MCU 90, and the hold switch 208 receives the switch control signal &quot; HOLD &quot;. Accordingly, the initial coefficients fed back from the initial coefficient data updating unit 214 are selectively provided to the initial coefficient updating unit 214 via the delay unit 212. The initial filter coefficients provided from the initial coefficient updating unit 214 to the FFE and DFE coefficient updating units 35 and 45 are used to receive the existing converged coefficient values to increase the initial convergence speed of the adaptive filter. That is, by using the existing converged value, it is possible to find and converge a stable filter coefficient more quickly.

Referring now to FIG. 6, a detailed block diagram of the highest order detector 60 in accordance with the present invention is shown. The highest order detector 60 includes a branch metric generator 310 for generating a branch metric, a branch metric comparator 330, a difference evaluation amount delay unit 400, and a survival path determiner 500.

The branch metric generator 310 subtracts the equalized channel signal in by using the reference level (+2, 0, -2) of the ideal equalized signal provided from the MCU 90 through the signal inverter 320. First through third subtractors 312 and 314 and first through third subtractions BM1, BM2, BM3, and BM4 by calculating absolute values of the outputs of the first and second subtractors 312 and 314. Absolute value circuits 322, 324, 326 are provided. The output of the first subtractor 312 is connected to the first absolute value circuit 322, the equalized channel signal value in is connected to the second absolute value circuit 324, and the output of the second subtractor 314 is connected to the first absolute value circuit 322. 3 is connected to the absolute value circuit 326.

The first absolute value circuit 322 generates as a metric BM3 having an absolute difference (| -2.0-in |) of the equalization channel signal with respect to the reference equalized signal level -2, and if the input signal level is an 8-bit value, ± 128 The video signal level is changed from ± 2 to ± 60. The second absolute value circuit 324 generates as a metric BM1, BM4 having an absolute difference (| 0.0-in |) of the equalization channel signal with respect to the reference equalization signal level 0, and the third absolute value circuit 326 has a reference equalization signal + Generate an absolute difference (| +2.0-in |) of the equalization channel signal for two as metric BM2. The branch metrics of each of the first to third absolute value circuits 322, 324, 326 are provided to the comparator 330. In an actual circuit, the reference equalization level is changed from 2.0 to a corresponding determination reference value according to the amplitude of the input signal, which is provided by the MCU 90.

The comparator 330 generates error metrics EM1 and EM2 for each state of the decoding grid using the branch metrics provided from the branch metric generator 310. For example, Equation 9 is a process of finding an error metric EM1 having a small error value among (BM1 + EM1c) and (BM3 + EM2c), and Equation 10 is a small error among (BM2 + EM1c) and (BM4 + EM2c). This process finds the error metric EM2 that has a value. This process can be expressed by the following equation.

EM1 = min ｛BM1 + EM1c, BM3 + EM2c｝

EM2 = min ｛BM2 + EM1c, BM4 + EM2c｝

Re-expression of Equations 9 and 10 is the same as Equations 11 and 12, respectively. That is, the determination of EM1 and EM2 is based on the judgment of the following equation.

BM1 + EM1c &lt; &Gt; BM3 + EM2c

BM2 + EM1c &lt; &Gt; BM4 + EM2c

Here, the variable for evaluating the difference between EM1 and EM2 (EM2-EM1) is defined as the difference evaluation amount (&quot; DEM &quot;) as follows.

DEM = EM2-EM1

DEMc = EM2c-EM1c

In order to determine the error metric EM1, Equation 7 is rewritten using Equation 9 as follows.

BM1-BM3 &lt; &Gt; EM2c-EM1c

BM1 &lt; &Gt; DEMc + BM3

In the same manner, to rewrite Equation 8 using Equation 9 to determine the error metric EM2, the following Equation 8 is obtained.

BM2-BM4 &lt; &Gt; EM2c-EM1c

BM2 &lt; &Gt; DEMc + BM4

In the comparator 330, the first adder 332 is used to calculate Equation 14, and an output from the first adder 332 is provided to the first selector 336 and the first comparator 342. The first comparator 342 compares the output of the first adder 332 with the output of the second absolute value circuit 324 and compares the result with the first selector 336 and the next stage survival path storage / update module 500. Output as a multiplexer control signal SS1.

Meanwhile, the second adder 334 is used to calculate Equation 15, and the output of the second adder 334 is provided to the second selector 338 and the second comparator 344. The output signals of the second comparators 342 and 344 are provided as multiplexer control signals SS2 of the second selector 338 and the survival path storage / update module 500 of the next stage.

In the operation, when the output BM1 of the second absolute value circuit 324 in the first comparator 342 is smaller than the output DEMc + BM3 of the first adder 332, the control signal SS1 is &quot; 0 &quot; If (BM1) is greater than (DEMc + BM3), control signal SS1 is output as &quot; 1 &quot;. Similarly, if the output BM2 of the third absolute value circuit 326 in the second comparator 344 is smaller than the output DEMc + BM4 of the second adder 334, the control signal SS2 is output as &quot; 0 &quot; If, (BM2) is greater than (DEMc + BM4), the control signal SS2 is output as &quot; 1 &quot;.

By the control signals SS1 and SS2 of the first and second comparators 342 and 344, the outputs of the first and second selectors 342 and 344 are input to the subtractor 350 so that the difference evaluation amount DEM, i.e. (EM2-EM1) Create a value. The output of the subtractor 350 is provided to the difference evaluation amount delay unit 400 of FIG. 7 and the survival path determiner 500 of FIG. 8.

Referring to Fig. 7, a detailed configuration of the difference evaluation amount delay unit is shown.

In the difference estimator delay unit 400, the second input terminal of the multiplexer 420 is connected to the output of the subtracter 350 and the output of the (1-D) delayer 426 provided through the line 356, and the multiplexer A first input of 430 is connected to the output of subtractor 350 and the output of delayer 436. An output of the multiplexer 420 is connected to a first input of an AND gate 422. The second input of AND gate 422 is connected to the clear signal CLR and its output is connected to the second input terminal of the multiplexer 424. The output of the multiplexer 424 is connected to the delay 426. An output of the delayer 426 is coupled to a first input of the multiplexers 420, 424 and a first input of an output multiplexer 470.

Similarly, the output of multiplexer 430 is connected to the first input of AND gate 432. The second input of AND gate 432 is connected to the clear signal CLR and its output is connected to the first input of multiplexer 434. The output of the multiplexer 434 is coupled to the delayer 436. An output of the delayer 436 is connected to a second input terminal of the multiplexers 430 and 434 and a second input terminal of the output multiplexer 470.

Meanwhile, the signal HOLD and the signal INITIAL are connected to the two-input AND gate 410. The output of the AND gate 410 is connected to the control signal input of the multiplexer 414 through an inverter 412. The output of the multiplexer 414 is connected to the delayer 416, the output of the delayer 416 is connected to the second input terminal of the multiplexer 414 and the inverter 418, and the output of the inverter 418 is connected to the multiplexer ( It is connected to the first input of 414 and is used as a control signal of the multiplexers 420, 430, and 470.

In this operation, the delayed symbol clock is compensated for one cycle in the asynchronous signal detection system by delaying the difference evaluation signal DEM provided from the subtractor 350 according to the HOLD signal according to the HOLD signal.

Referring now to FIG. 8, a detailed schematic diagram of survival path determiner 500 is shown. As shown, the survival path determiner 500 includes a survival path selection module 510 and a determination signal generator 560.

The survival path selection module 510 includes multiple stages of survival path storage / update units 420, 430, and 440 that determine survival paths in the first and second states. Each of the survival path storage / update units 420, 430, and 440 includes multiplexers 421, 431, 441 and path memories 423, 433, 443, which determine the survival path in the first state. And two-input multiplexers 422, 432, 442 and path memories 425, 434, 444 that determine the survival path of the second state. The survival path selection module 510 also includes path memories 425, 435 and 445 connected to outputs of the path memories 423, 433 and 443 in the first state and path memories 425, 434 and 444 in the second state, respectively. Path memory pairs 426, 436, and 446, respectively, connected to the output of the sub-head.

In the case of partial response IV, even- and odd-numbered code strings are affected only by even- and odd-numbered code strings, respectively, so that they are divided according to the Viterbi algorithm and finally combined by the multiplexer. Thus, path memory pairs 423, 425, 433, 435, 443, 445 in the first state and path memory pairs 424, 426, 434, 436, 444, 446 in the second state. Each path memory in i) can reduce the number of gates by using an even and odd multiplexer. The advantage of the present invention is thus used to alternately store and update the surviving paths of even and odd signals using path memory pairs, which may be configured as one shift register.

The control terminal of each of the multiplexers 421, 431, 441 that selects the survival path in the first state in the survival path selection module 510 is provided via the line 352 from the error metric generator 330 via the selection signal SS1. Is connected to a signal source providing signals 0 and -2 in a first state, and an input of the two-input multiplexer 431 is connected to path memories 425 and 426. The input of the two-input multiplexer 441 is connected to the output of the path memories 435 and 436. Similarly, the control terminal of each multiplexer 422, 432, 442 that selects the survival path in the second state in the survival path selection module 510 is provided via selection line SS2 provided from line 354 from the comparator 330. Is connected to a signal source providing signals 0 and +2 in a second state, and an input of the two-input multiplexer 432 is connected to the path memories 426 and 425. And the input of the two-input multiplexer 442 is connected to the output of the path memories 436 and 435.

The output of the path memories 445 and 446 of the surviving path storage / update unit 440 of the last stage is connected to the input of the multiplexer 440, and the control signal input terminal of the multiplexer 450 is a bit code through the line 370. It is connected to the output of the detector 360.

The sign bit detector 360 performs a function of determining the magnitude of the difference evaluation amount DEM as a sign. In other words, the data output with the two's complement value is negative if the MSB of the data is &quot; 1 &quot; and positive if &quot; 0 &quot;. That is, if the difference evaluation amount DEM is greater than zero, the sign bit detector 360 outputs zero as sign bits, and if the difference evaluation amount DEM is less than zero, the sign bit detector 360 outputs one as sign bits. This sign bit is provided as a selection control signal that controls multiplexer 450 via line 358.

The operation of the survival path selection module 500 will be described, for example, with respect to an odd number of signals reproduced from the head drum.

In FIG. 10A, a state diagram generated by survival path determiner 500 is shown. As can be seen in this figure, the selection signals SS1 and SS2 generated at each symbol time by the survival path determiner 500 are shown in the following table.

t1 t2 t3 ... t4 SS1 One 0 One ... 0 SS2 0 One 0 ... One

Path memory value at symbol time t1 Memory (423/424) Memory (433/434) ... Memory (443/444) State 1 -2 0 ... 0 State 2 0 0 ... 0

Path memory value at symbol time t2 Memory (423/424) Memory (433/434) ... Memory (443/444) State 1 0 -2 ... 0 State 2 +2 -2 ... 0

Path memory value at symbol time t3 Memory (423/424) Memory (433/434) ... Memory (443/444) State 1 -2 +2 ... 0 State 2 0 +2 ... 0

Path memory value at symbol time t4 Memory (423/424) Memory (433/434) ... Memory (443/444) State 1 0 -2 ... +2 State 2 +2 -2 ... +2

As such, the survival path converges while the path that does not lead through the path memory is deleted. In FIG. 10B, the path of the dotted line survives and the path of the solid line is deleted. Deleting a survival path means that two path memories are filled with only surviving values.

As described above, the same operation is performed on the even-numbered signal reproduced from the head drum, and detailed description thereof is omitted.

As a result, the multiplexer 450 selectively converts the survival code values (± 2, 0) provided from the path memories 445 and 446 in the first and second states according to the control signal from the sign bit detector 370 into &quot; ± 2 &quot; is 1 and &quot; 0 &quot; is &quot; 0 &quot; Outputting as will output the detected bits.

On the other hand, in the survival signal output unit 560, the output of the multiplexer 462 is connected to the delay unit 464, the output of the delay unit 464 is connected to the multiplexer 466, the output of the multiplexer 466 is Connected to a retarder 467. Multiplexers 462 and 466 are controlled by a control signal HOLD provided over line 358.

The output of the path memories 423, 424 is connected to the multiplexer 470, the output of the path memories 425, 426 is connected to the multiplexer 480, and the output of the path memories 433, 434 is the multiplexer 490. Is connected to. Multiplexers 470 and 490 are controlled by the output of delayer 464, and multiplexer 480 is configured to be controlled by the output of delayer 467.

The outputs of the multiplexers 470, 480, 490 are provided to the phase detector 70 (see FIG. 2), respectively.

As described above, according to the present invention, it is possible to design the equalizer in a simpler configuration rather than in the configuration of a full bit adder, and the advantage of more accurate detection of the source signal from the signal reproduced from the magnetic channel is provided.

Claims (21)

An asynchronous partial response channel data detection apparatus for estimating code data recorded in the transmission channel from digital data sampled from code data reproduced from a transmission channel, Adaptive equalization means for outputting the digital data as an equalization signal having partial response channel characteristics according to adaptive filter coefficient values; Phase control means (100) for synchronizing the sampling time of the digital data and the code time difference of the sign data by detecting and compensating for a phase error using the signal output from the equalizing means; The level of the ideal equalization signal of the first and second states, respectively selected according to the first and second selection signals, is determined as the sign data for the equalized signal, wherein the selected sign data is the first and second previous signals. And a best-order detector, wherein each state starts from the two possible states of and each state survives in such a way that it transitions to one of the two states. The method of claim 1, wherein the highest order detector, A branch metric generator 310 for generating a difference between each level of the reference equalized signal and the equalized signal as a branch metric representing possible states for the respective states; The first selection signal for determining the magnitude of the difference between the two metrics of the first state with respect to the difference evaluation amount and selecting the branch metric having the smaller difference therein as the current error metric and the second for the difference evaluation amount A comparator 330 for determining the magnitude of the difference between the two metrics of the state and generating the second selection signal for selecting the branch metric having the smaller difference therein as the current error metric; A difference evaluation amount generator (350) for generating a difference between the current error metrics for the first and second states selected by the comparison unit as the difference evaluation amount; And a survival path determiner for tracking a survival path of the survival code data according to the first and second selection signals generated from the comparator. The method of claim 2, wherein the branch metric generator, A first absolute value circuit (322) for generating as a matrix BM3 a difference (| -2.0-in |) of the equalized signal with respect to the level -2 of the reference equalized signal; A second absolute value circuit (324) for generating the difference (| 0.0-in |) of the equalized signal to level 0 of the reference equalized signal as the metrics BM1, BM3; And a third absolute value circuit 326 for generating as a matrix BM2 the difference (| +2.0-in |) of the equalized channel signal with respect to +2 of the reference equalized signal. Detector. The method of claim 2, wherein the comparison unit, A first adder (332) for adding the output of the first absolute value circuit (322) and the difference evaluation amount; A second adder (334) for adding the output of the second absolute value circuit (324) and the difference evaluation amount; A first comparator (342) for generating the first selection signal by comparing the output of the first adder (332) with the output of the second absolute value circuit (324); A second comparator (344) for generating the second select signal by comparing the output of the second adder (334) with the output of the third absolute value circuit (324); A first selector (336) for selectively outputting the output of the first adder (332) and the output of the second absolute value circuit (324) to the difference estimation amount generator (350) according to the first selection signal; And a second selector 336 for selectively outputting the output of the second adder 332 and the output of the third absolute value circuit 324 to the difference estimation amount generator 350 according to the second selection signal. Asynchronous partial response channel data detector. The method of claim 2, wherein the survival path determiner 500 A survival path selection module (510) having multiple stages of survival path storage / update units (520, 530, 540) for storing surviving code data of the first and second states; And a determination signal generator (560) for determining a survival path of the survival path selection module (510). The multiplexer 421 of claim 5, wherein each of the multi-stage survival path storage / update units 420, 430, and 440 determines the code data surviving in the first state in response to the first selection signal. 431, 441 and path memories 423, 433, 443 for storing and updating survival code data determined by the respective multiplexers; A path memory for storing and updating the multiplexers 422, 432, and 442 for determining the code data surviving in the second state in response to the second selection signal and the survival paths determined by the respective multiplexers; 425), (434), (444); Selectively outputs survival code data provided from the path memories 443 and 443 of the first and second states of the survival path storage / update unit 440 at the last stage as the nearest code data according to a code data selection control signal. A sign data selector 450; And a code bit detector (360) for judging whether the difference evaluation amount (DEM) is greater than or less than zero and outputting the coded data selection control signal. 3. The multiplexer 522 of claim 2, wherein the input of the multiplexer 421 in the first stage of the top stage is connected to a signal source providing levels 0 and -2 of the reference equalization signal, and the multiplexer 522 in the second stage of the top stage. And an input of is connected to a signal source providing a level 0 and a +2 of the reference equalized signal. 8. The asynchronous partial response channel data detection device of claim 7, wherein the survival signal is provided from at least two stages of the survival path memories in the first and second states in the survival path storage / update unit. 3. The method according to claim 2, wherein the highest order determiner further comprises a delay means (400) for delaying the difference evaluation amount signal DEM provided from the difference evaluation amount generating means (350) for a preset period according to the delay control signal (HOLD). Asynchronous partial response channel data detection device. 6. The apparatus of claim 5, wherein the transmission channel comprises a magnetic recording medium. 11. The method of claim 10, wherein the survival path selection module 510 is a path of the second state and the path memory (425, 435, 445) respectively connected to the output of the path memory (423, 433, 443) of the first state Further comprising path memories 426, 436, 446 connected to the outputs of the memories 425, 434, 444, respectively, to form a memory pair, Path memory pairs 423, 425, 433, 435, 443, 445 of the first state and path memory pairs 424, 426, 434, 436, 444, 446 of the second state Wherein each path memory is used to alternately store and update surviving paths for signals reproduced alternately from the magnetic recording medium. 12. The apparatus of claim 11, wherein the path memory pairs in the first and second states are configured as shift registers. The method of claim 1, wherein the equalizing means, A first filter coefficient updater 35 providing a first filter coefficient; A second filter coefficient updater 45 for providing a second filter coefficient; A feedforward equalizer 30 which multiplies the first filter coefficient by the equalization signal to form a signal having the partial response channel characteristic; A decision feedback equalizer 40 which multiplies the output of the highest order detector by a second filter coefficient to remove inter-sign interference of the symbol data; An adder (50) for adding the output of the feed forward equalizer and the output of the decision feedback equalizer to provide the equalized signal to the highest order detector; And an error signal generator (55) for generating a difference between an output of said adder and an output of said highest order detector as an error signal. The filter coefficients of claim 13, wherein the filter coefficients of the first and second filter coefficient update units respectively satisfy the following equations,

In the above equation, 1 is a convergence constant, e is an error signal from the error signal generator, X is an output signal of the highest order detector, and C is a filter coefficient. 14. The asynchronous part of claim 13, wherein the first and second filter coefficient updaters each include an initial data updater 214 for providing initial filter coefficients to the feedforward equalizer and the decision feedback equalizer. Response channel data detector. According to claim 1, wherein the phase control means is provided with a phase detection unit 70 for outputting a phase error signal Zk as shown in the following equation by using a direct decision (decision directed) method,

Wherein e is an error signal from the error signal generator (50) and X is a survival signal provided from the highest order detector. 17. The asynchronous partial response channel data detector of claim 16, wherein said phase control means further comprises a loop filter (80) for removing high frequency components or noise from the output of said phase detector (70). The method of claim 17, wherein the output of the loop filter 80 satisfies the following equation,

1 _k is an output of a loop filter (80), α and β are filter constants, and Zk is a phase error signal from the phase detector (70). The filter of claim 1, wherein the interpolation filter is A first adder (810) for adding the phase error signal to W which is the ratio of the sampling frequency to the actual recording data frequency; An accumulating module 840 accumulating the output of the first adder 810; A subtractor 860 for generating a signal 1k representing the distance between the sampled data Xn and the actual recorded data from the accumulated value provided from the accumulation module 840 as an address signal for reading the filter coefficient storage area of the lookup table. ; An ideal response characteristic curve of an impulse response filter has a plurality of filter coefficient storage areas for storing response values corresponding to each sampling time as the filter coefficients, and the signal 1k representing the distance is an address signal for reading the filter coefficient storage area. A lookup table 910 used as; A finite response filter that delays the survival data signal, multiplies the delayed survival data signal by the filter coefficients read from the lookup table 910, and adds the multiplied values to output the compensated signal to the equalization means. And a unit (920) for asynchronous partial response channel data detection. 20. The asynchronous partial response channel data detector of claim 19, wherein the accumulation module outputs the delay control signal &quot; HOLD &quot; when the accumulated value reaches a predetermined period. 20. The filter of claim 19, wherein the interpolation filter is An asynchronous partial response channel data detector comprising a finite impulse response filter.

Images