JPH05101306A - Magnetic reproducing device - Google Patents

Abstract

PURPOSE:To quicken the convergence of a filter coefficient and to obtain an excellent stable stage. CONSTITUTION:After the characteristic of a regenerative signal from a magnetic head 11 is compensated by a filter 14 to be an equalizer, the signal is demodulated by a demodulation circuit 15. By an adaptive control part 17, the filter coefficient of the filter 14 is controlled to revise based on a demodulation error (residual difference) in the demodulation circuit 15 and an input to the filter 14. The revised amount is specified by a revision coefficient mu. By a revision coefficient control part 18, the revision coefficient muis controlled in accordance with the convergence condition of the filter coefficient of the filter 14 controlled by the adaptive control part 17.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a magnetic reproducing device,
In particular, the present invention relates to a magnetic reproducing device for converting a video signal into a digital signal and reproducing a signal recorded on a magnetic recording medium by utilizing a so-called partial response system.

[0002]

2. Description of the Related Art Generally, in magnetic recording / reproducing, an equalizer is used to compensate for amplitude distortion, phase distortion, etc. due to recording / reproducing characteristics with respect to a magnetic recording medium. In recent years, even in such magnetic recording / reproduction, an adaptive equalization method used in communication has been adopted.

Adaptive equalization has been conventionally developed as a technique for performing high-speed data transmission using a telephone line. In a telephone line, the transmission line characteristics change depending on the connection state of the line. Therefore, if the fixed equalizer is used, the transmission line characteristics cannot be completely corrected, and it becomes necessary to adaptively adjust the characteristics of the equalizer.

In such a communication system, a method (automatic equalization) of transmitting a necessary signal after transmitting a signal whose waveform is known in advance and checking transmission line characteristics, and a signal to be transmitted itself are used. There is a method (adaptive equalization) for examining transmission line characteristics. In either case, the purpose of the equalizer is to faithfully restore the transmission signal waveform by automatically removing the distortion from the reception signal waveform that has been distorted when passing through the transmission path.

In order to consider the application of the above-mentioned adaptive equalization to magnetic recording / reproducing, a magnetic tape (video tape) is used as a magnetic recording / reproducing apparatus by converting a video signal into a digital signal and using a so-called partial response system. It is assumed that a digital VTR (video tape recorder) that records and reproduces data is recorded. The partial response method is a method of shaping the spectrum of the code by positively utilizing intersymbol interference due to the transfer characteristic of the transmission path (or recording medium). For example, in the partial response class IV, NRZI code, interleaved NR
ZI code and the like belong. A so-called precoder is provided on the recording side, and the input data is converted into an intermediate sequence in order to avoid the propagation of code errors during reproduction (discrimination). FIG. 6 shows an example of the configuration on the reproducing side when the adaptive equalization system is adopted in a digital VTR that performs magnetic recording and reproduction by using such a partial response system.

In FIG. 6, a magnetic signal recorded on a magnetic tape (not shown) is converted into an electric signal by the magnetic head 101, amplified by the reproduction amplifier 102, and sent to the detection characteristic circuit 103. The detection characteristic circuit 103 has a characteristic (1 + D) which is the detection characteristic (encoding characteristic) of the partial response.
The output signal from the detection characteristic circuit 103 is a so-called FIR.
After being supplied to the equalizer 104 including a (finite impulse response) filter or a transversal filter and subjected to an adaptive equalizing process, the decoding circuit 10
5, the data sequence is decoded at the time of recording by discriminating between "1" and "0" by level comparison (comparing) or the like.

The output d from the decoding circuit 105 is sent to the adder (error detector) 106 and the output y from the equalizer 104 is subtracted, whereby an error (residual) e is taken out, and this error e is obtained. It is sent to the adaptive control unit 107. The output x from the detection characteristic circuit 103 is supplied to the adaptive control unit 107 as a so-called reference input. The adaptive control unit 107 adjusts the filter characteristic of the equalizer 104 so as to minimize the signal power of the error (residual error). When a so-called transversal filter is used for the equalizer 104, the multiplication coefficient (tap coefficient) for each tap is adaptively modified and updated, so that the characteristics of the transversal filter are the electromagnetic characteristics during magnetic recording and reproduction. It is adjusted to have a shape close to the inverse characteristic of the conversion characteristic.

The output from the decoding circuit 105 is sent to the signal processing circuit 108 for reproduction of the sync block, error correction, etc., and then to the video signal processing circuit 109 for restoration of the original image data. .. Besides this, although not shown,
The output data from the signal processing circuit 108 is sent to an audio signal processing circuit, a subcode signal processing circuit, etc., and each processing is performed.

[0009]

The update of the multiplication coefficient (tap coefficient) for each tap of the transversal filter used in the equalizer 104 is generally given by the following equation. W _{k + 1} = W _k +2 με _k X _{k In} this formula, W is a tap coefficient vector, μ is a change coefficient, ε is an error signal, X is an input signal, and subscripts k and k + 1 are input data sequences, respectively. The kth sample and the (k + 1) th sample are shown. Here, the change coefficient (also referred to as a convergence coefficient) μ is a coefficient that defines the amount of change for one time, and affects the convergence speed of the coefficient and the stability after convergence. When the value of μ is large, the convergence is fast, but the fluctuation is large for a small error, so that it becomes difficult to obtain a good stable state. On the other hand, if μ is too small, it takes too much time to converge, which is not practical. Conventionally, the value of μ has been determined in consideration of the two, and it has been difficult for both to achieve excellent characteristics.

The present invention has been made in view of such a situation, and the tap coefficient converges quickly when controlling the multiplication coefficient (tap coefficient) for each tap of the digital filter which is the adaptive equalizer. It is also an object of the present invention to provide a magnetic reproducing device that can obtain a good stable state.

[0011]

A magnetic reproducing device according to the present invention is a magnetic reproducing device for reproducing magnetic data recorded on a magnetic recording medium, and is a filter serving as an equalizer for compensating for characteristics of a reproduced signal from a magnetic head. A decoding circuit that decodes an output signal from the filter; an adaptive control unit that adaptively adjusts the characteristics of the filter based on an input signal to the filter and an input / output signal to the decoding circuit; 1 of the filter coefficient W that determines the characteristics of
The above-mentioned problem is solved by having a change coefficient control unit that controls the value of the change coefficient μ that defines the change amount of the number of times according to the convergence state of the filter coefficient W.

Here, as a criterion for switching the change coefficient μ, an error ε between the input and output signals for the decoding circuit,
It is preferable to use at least one of ε ² , the amount of change Δε, Δε ² , and the amount of change ΔW, ΔW ² of the filter coefficient W.

[0013]

The function of changing the coefficient μ according to the convergence state of the filter coefficient W
Is changed to a large μ value in the initial state, and is switched to a small μ value in a state where the convergence is advanced, so that the convergence is fast and a good stable state can be obtained.

[0014]

1 is a block circuit diagram showing a schematic structure of a reproducing system of a digital VTR as described above as an embodiment of a magnetic reproducing apparatus according to the present invention. In FIG. 1, a magnetic signal recorded on a magnetic tape (not shown) is converted into an electric signal by a magnetic head 11 in a mechanical block (not shown) of a VTR and then amplified by a reproduction amplifier 12. It is sent to the detection characteristic circuit 13. The detection characteristic circuit 13 has the characteristic (1 + D) which is the detection characteristic (encoding characteristic) of the partial response described above. The output signal from the detection characteristic circuit 13 is supplied to the filter 14, which is the main part of the equalizer. A so-called FIR (finite impulse response) filter or a transversal filter is generally used as the filter 14, and its filter characteristic is adaptively adjusted by an adaptive control unit 17 described later. The output signal from the filter 14 is supplied to the decoding circuit 15 and "1" by level comparison (comparing) or the like,
The determination of "0" is made, and the data series at the time of recording is decoded. The output signal from the decoding circuit 15 is taken out via the output terminal 15 _OT and sent to the signal processing circuit 108 shown in FIG.

The adder (error detector) 16 is the decoding circuit 1
An error (residual) e is taken out by subtracting the output y of the equalizer filter 14 from the output d of 5, and this error e is sent to the adaptive control unit 17. The output x from the detection characteristic circuit 13 is supplied to the adaptive control unit 17 as a so-called reference input. The adaptive control unit 17 causes the filter 14 to minimize the signal power of the error (residual error).
By adjusting and updating the coefficient (tap coefficient) of, the equalizer characteristic is adjusted so as to have a form close to the inverse characteristic of the electromagnetic conversion characteristic during magnetic recording and reproduction. That is, filter 1
4 and the adaptive control unit 17 constitute a so-called adaptive filter, and the adaptive equalizer can be regarded as one using an adaptive filter for the equalizer.

Next, the change coefficient control unit 18 of FIG. 1 controls the change coefficient μ of the filter coefficient according to the convergence state of the filter coefficient of the filter 14, and more specifically,
For example, some preset change coefficients μ are switching-controlled according to the square of the error e ² . In the example of FIG. 3 described later, μ = 0.1 is set to a large value in the initial state, and μ = μ when the error squared ε ² reaches a predetermined value ε _sw ^2.
Switching control is performed to a small value of 0.01. In addition to e ² , the error amount e and error changes Δe and Δe ² can be used as the criterion for the convergence state. The change amount [Delta] w _i of each filter tap coefficients w _i, and the square of [Delta] w _i ^2,
These total amounts ΣΔw _i , ΣΔw _i ^{2 and the} like can also be used as criteria for determining the convergence state of the filter coefficient. The above-mentioned e ² , e, Δe, Δe ² , Δw _i , Δ which are the criteria for judging these
w _i ² , ΣΔw _i , ΣΔw _i ^2, etc. are the adaptive control unit 1 described above.
7, any of these may be used alone or in combination.

As described above, by controlling the change coefficient μ according to the convergence state of the filter coefficient, it is possible to realize a more stable convergence state while increasing the convergence speed of the filter coefficient.

Next, the filter 14 and the adaptive controller 17
An example of a specific configuration of a so-called adaptive filter including and will be described with reference to FIG. In FIG. 2, the reference input x from the input terminal 14 _IN is a delay element corresponding to the number of taps, for example, four delay blocks 21a and 21.
It is sent to the series circuit of b, 21c and 21d. Input x ₀ from the input terminal 14 _IN and each delay element 21a, 21
Outputs x- ₁ , x- ₂ , x- ₃ , x from b, 21c, 21d
_-4 is a coefficient multiplier 22a, 22b, 22c, 2 respectively
2d and 22e, and are multiplied by filter coefficients (filter tap coefficients) w ₀ , w ₁ , w ₂ , w ₃ and w ₄ , respectively, and then added. That is, the coefficient multipliers 22a, 2
The outputs from 2b are added by the adder 23a, the output from the coefficient multiplier 22c and the output from the adder 23a are added by the adder 23b, and the same applies to the coefficient multipliers 22d and 23d in the adders 23c and 23d. The output from 22e is also sequentially added to form an output y, which is sent to the decoding circuit 15. Each filter coefficient w ₀ , w ₁ , w ₂ , w ₃ , w ₄ is
It is adapted to be modified by a coefficient modification (update) control signal from the adaptive controller 17.

As the adaptive algorithm used in the adaptive control section 17, many methods have been proposed.
As one specific example thereof, an LMS (least mean square, least mean square) algorithm will be described. Here, the number of the delay elements is generalized to L, and the delay elements 21 ₁ , 21 ₂ , ..., 21 _L are set. At this time, the first input x ₀ and each of these delay elements 23 ₁ ,
23 ₂ , ..., Each output from 23 _L x _-1 , x _-2 , ...
., X- _L are coefficient multipliers 24 ₀ , 24 ₁ , 2 respectively
4 _2, ..., is transmitted to the 24 _L, the filter coefficients are w _0, w _1, w _2, is multiplied ..., and w _L, shall be added is sent to the adder.

The input data at the k-th sampling period (time k) of the data sequence of the input x and the delayed output data from the delay elements 21 ₁ , 21 ₂ , ..., 21 _L are respectively x _k , X _k-1 , x _k-2 , ..., x
_{When kL} is set, the input vector X _k subjected to FIR filtering is set as X _k = [x _k x _k-1 x _k-2 ... X _kL ] ^T (1). T in the equation (1) represents a transposed symbol. For this input vector X _k, each filter coefficient (weighting coefficient) and _{w k0, w k1, w k2} , ···, w kL, when the FIR filter output and y _k, the relationship between the input and output follows (2)
It becomes like a formula.

Y _k = w _k0 x _k + w _k1 x _k-1 + ... + w _kL x _kL (2) Further, a filter coefficient vector (weight vector) W
a _k, if W _k = _{[w k0 w k1 w k2 ··· w} kL ] ^T ··· (3) the definition, input-output relationship ^{_{is, y k = X k T W}} k = W k T X k ... is described as (4). If the desired response is d _k , the error ε _{k from the} output is expressed as ε _k = d _k −y _k = d _k −X _k ^T W _k (5). The following equation is used to update W _k so that ε _k approaches 0.

[0022] W _{k +} 1 = is _{W k -μ ▽ k ··· (6} ) μ in this equation, is the gain factor that determines the speed and stability of adaptation, ▽ _k represents the gradient. The LMS algorithm, ▽ _k is not estimated from the short-term average of epsilon _k ^2, used by partially differentiating epsilon _k ² directly. ^{_{▽ k = δε k 2 / δW}} = -2ε k X k ··· (7) The equation (7) is substituted into the equation (6), the coefficient update _{equation, W k + 1 = W k} + 2με k It is expressed as X _k (8).

In the filter tap coefficient updating equation of the above equation (8), the speed of convergence of the filter coefficient differs depending on the value of the change coefficient μ, and the error amount in the converged state also differs. .. That is, in FIG. 3, the change coefficient μ = 0.
The change of the square ε ² of the error with the passage of time when 01 (broken line) and when μ = 0.1 (dashed line) is shown. As is apparent from FIG. 3, when the change coefficient μ is large as 0.1, the convergence speed can be increased, but ε ² in the converged state is large, and the stability after convergence is not good. When the coefficient μ is as small as 0.01, ε ² in the finally converged state can be made small, but there is a drawback that the convergence speed is slow and it takes too long to converge.

On the other hand, the solid line in FIG. 3 is one specific example of the embodiment of the present invention. In the initial state, the change coefficient μ is set to a large value of 0.1, and the convergence progresses to a predetermined convergence state. Once t _sw for example epsilon ² has reached the predetermined value ε _sw ^2, μ
The switching is controlled to a small value of 0.01. As is clear from the concrete example shown by the solid line in FIG. 3, by changing the value of μ from 0.1 to 0.01 in accordance with the convergence state, the convergence is made faster than in the case where a fixed value is used. However, a state in which the error amount is smaller can be realized.

In addition to the above ε ² , the error amount ε and the changes Δε and Δε ² thereof can be used as the criterion for the convergence state, and the changes Δw _i and Δw of the filter tap coefficients can be used.
_i ² and, the sum of these amounts Shigumaderutadaburyu _i, are as described above can be utilized Shigumaderutadaburyu _i ² like. Further, in the specific example of FIG. 3 described above, the change coefficient μ is switched in two steps, but
It is similarly effective to switch to a multi-stage of more than one stage. It is also preferable that the change coefficient μ be continuously changed as a continuous function of ε.

In addition to this, the present invention is not limited to the above-mentioned embodiment, and for example, the specific configuration of the filter 14 and the algorithm used in the adaptive control unit 17 are the FIR filter and the LMS algorithm of the above-mentioned embodiment. Not limited. Further, the applicable equipment is not limited to the digital VTR, and the present invention can be applied to a digital tape recorder, an analog VTR and the like.

[0027]

As is clear from the above description, according to the magnetic reproducing apparatus of the present invention, in the magnetic reproducing apparatus for reproducing the magnetic data recorded on the magnetic recording medium, the reproduction signal from the magnetic head is reproduced. An adaptive filter is used as an equalizer for compensating the characteristics, and the value of the change coefficient μ that defines the amount of one change of the filter coefficient W that determines the characteristics of this filter is controlled according to the convergence state of the filter coefficient W. , The filter coefficient converges quickly, and a good stable state can be obtained.

Specifically, the change coefficient μ is switch-controlled according to the convergence state of the filter coefficient W, for example, a large μ value is set in the initial state, and then it is changed to a small μ value as the convergence progresses. The error in the converged state can be reduced while the convergence is accelerated.

[Brief description of drawings]

FIG. 1 is a block circuit diagram showing a schematic configuration of a part of a reproducing system of a digital VTR as an embodiment of a magnetic reproducing apparatus according to the present invention.

FIG. 2 is a block circuit diagram showing a specific example of an internal configuration of an adaptive equalizer (adaptive filter) used in the embodiment.

FIG. 3 is a graph showing a temporal change of an error ε ² according to a value of a change coefficient μ.

FIG. 4 is a block circuit diagram showing a schematic configuration of a reproduction system of a digital VTR used for explaining a conventional technique.

[Explanation of symbols]

 11 ... Magnetic head 13 ... Detection characteristic circuit 14 ... Filter (equalizer) 15 ... Decoding circuit (comparator) 16 ... Adder (error detector) ) 17 ... Adaptive control unit 18 ... Change coefficient control unit

Claims (2)

[Claims] 1. A magnetic reproducing apparatus for reproducing magnetic data recorded on a magnetic recording medium, a filter serving as an equalizer for compensating characteristics of a reproduced signal from a magnetic head, and a decoding circuit for decoding an output signal from the filter. An adaptive control unit that adaptively adjusts the characteristics of the filter based on an input signal to the filter and an input / output signal to the decoding circuit, and a single change of the filter coefficient W that determines the characteristics of the filter. The value of the change coefficient μ that defines the amount is set to the filter coefficient W
And a change coefficient control means for controlling the magnetic reproduction apparatus according to the convergence state of the magnetic reproducing apparatus. 2. As a reference for controlling the value of the change coefficient μ, an error ε between the input and output signals with respect to the decoding circuit,
The magnetic reproducing apparatus according to claim 1, wherein at least one of ε ² , changes Δε and Δε ² thereof, and changes ΔW and ΔW ² of the filter coefficient W is used.