JPH097300A - Equalizer and magnetic recording / reproducing device - Google Patents

Abstract

(57) [Summary] [Object] To provide an equalizer capable of canceling different waveform distortions with a simple circuit configuration even when a conventional linear signal processing method is used. In an equalizer that compensates for waveform distortion in an input signal, the input signal value is converted into an Nth-order function conversion characteristic (N> 1).
Amplitude value converter to convert to another value by, a non-recursive filter to equalize the output of the amplitude value converter, the size of the equalization error of the output of the non-recursive filter as a criterion, A learning unit that learns and updates the coefficient that determines the characteristic of the amplitude value converter and the tap coefficient of the acyclic filter so that the equalization error is minimized.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an equalizer and a magnetic recording / reproducing apparatus, and more particularly to a technique effective when applied to a magnetic recording / reproducing apparatus using an MR reproducing head.

[0002]

2. Description of the Related Art In a magnetic recording / reproducing apparatus such as a magnetic disk apparatus, the sensitivity of a reproducing head is being increased in order to increase the density, and a reproducing head using a magnetoresistive effect (instead of a conventional inductive head is used). Hereinafter, this is referred to as an MR head) has been developed.

FIG. 9 is a diagram showing the magnetic field-reproduction output conversion characteristic 76 of the MR head, and is a diagram for explaining the nonlinearity of the magnetic field-reproduction output conversion characteristic of the MR head.

As shown in FIG. 9, in a reproducing system of a magnetic recording / reproducing apparatus using an MR head, a bias magnetic field 73 is superposed on an input magnetic field 71 from medium magnetization of a recording medium,
An operation area is set in a linear portion on the magnetic field-reproduction output conversion characteristic 76 to reproduce the signal recorded on the magnetic recording medium.

In the reproducing system of the magnetic recording / reproducing apparatus using the MR head, the deviation of the bias magnetic field 73 and M
Due to variations in the magnetic characteristics of the R element, the magnetic field-reproduction output conversion characteristic 76 of the MR head becomes non-linear, and waveform distortion represented by up-down asymmetry of the reproduction signal waveform occurs in the reproduction signal reproduced by the MR head.

On the other hand, recently, in magnetic recording / reproducing devices such as hard disk devices and digital VTRs, PRML (par)
tial response maximum like
Signal processing called "lihood" has been attracting attention.

In the PRML signal processing, the recording density is 1.2 by the signal processing without largely changing the existing recording / reproducing system.
It is a technology to increase the amount by 1.5 times.

FIG. 10 is a diagram for explaining the outline of the PRML signal processing.

As shown in FIG. 10, the PRML signal processing means the recording encoder 601 and the precoder 60 based on an input information sequence from a hard disk controller (HDC) or the like.
2. A write signal to a recording medium such as a hard disk device is generated via the write compensation circuit 603, and an automatic gain control circuit (A
GC) 604, low-pass filter 605, A / D converter 606, PR (partial response) equalizer 607,
Via the Viterbi decoding circuit 608 and the recording decoder 609,
It is for reproducing an output information sequence to a hard disk controller (HDC) or the like.

In FIG. 10, the servo signal decoding circuit, the timing control circuit of the A / D converter 606 and the like are omitted.

Generally, in a magnetic recording / reproducing apparatus, a signal processing system such as the PRML signal processing is designed on the assumption that the magnetic field-reproduction output conversion characteristic of the magnetic recording / reproducing system is linear. The non-linearity of the magnetic field-reproduction output conversion characteristic of the system is a major cause of performance deterioration of the magnetic recording / reproducing apparatus.

In the conventional magnetic recording / reproducing apparatus, the PR
The transversal type filter was used for the equalizer to equalize the waveform distortion in the reproduction signal due to the non-linearity of the magnetic field-reproduction output conversion characteristic of the magnetic recording / reproduction system. It was difficult to sufficiently equalize the waveform distortion in the reproduced signal due to the non-linearity of the output conversion characteristic.

Therefore, in the preceding stage of a waveform equalizer such as a PR equalizer, the distortion is reduced by canceling the waveform distortion in the reproduced signal due to the non-linearity of the magnetic field-reproduction output conversion characteristic of the magnetic recording / reproduction system. The method of doing is proposed.

For example, the following publication (a) describes a method of A / D converting a reproduced signal and then converting an amplitude value using a lookup table.

In this case, the magnetic field-reproduction output conversion characteristic of each MR head is measured in advance by changing the bias magnetic field of the MR head, and when the operating point on the magnetic field-reproduction output conversion characteristic changes. It is a method of performing fitting based on a stored table.

(A) Japanese Patent Application Laid-Open No. 6-44510 (Adaptive digital linearization method and device for magnetoresistive head output) Further, a RAM (Random Ac) is used as a decision feedback equalizer.
A method for dealing with non-linearity such as non-linear bit shift has been proposed by using an equalizer incorporating cess Memory).

[0017]

In the magnetic recording / reproducing system, in the preceding stage of the waveform equalizer such as the PR equalizer, the reproduced signal in the reproduced signal caused by the non-linearity of the magnetic field-reproduced output conversion characteristic of the magnetic recording / reproduce system. The method of canceling the waveform distortion is an effective method.

However, the non-linearity of the magnetic field-reproduction output conversion characteristic of the MR head has different characteristics for each MR head, and in order to correspond to this, the waveform distortion in the reproduction signal reproduced by the MR head is measured. However, it is necessary to set the characteristics of the device to be canceled by some kind of learning.

It is relatively easy to measure the waveform distortion in the reproduced signal with the MR element alone, but it is very difficult to measure the waveform processed in the head.

The method described in the above-mentioned publication (JP-A-6-44510) requires a circuit for arbitrarily setting the bias magnetic field and a dedicated learning circuit.

The present invention has been made in order to solve the above-mentioned problems of the prior art, and an object of the present invention is to simplify the equalizer even when the conventional linear signal processing method is used. It is an object of the present invention to provide a technique capable of canceling different waveform distortions with a circuit configuration.

Another object of the present invention is, in a magnetic recording / reproducing apparatus using an MR head as a reproducing head, even if a conventional linear signal processing system is used, the MR head has a simple circuit configuration and differs from MR head to MR head. It is an object of the present invention to provide a technique capable of canceling waveform distortion caused by non-linearity of magnetic field-reproduction output conversion characteristics.

The above and other objects and novel features of the present invention will become apparent from the description of the present specification and the accompanying drawings.

[0024]

SUMMARY OF THE INVENTION Among the inventions disclosed in the present application, the outline of a representative one will be briefly described.
It is as follows.

(1) In an equalizer for compensating for waveform distortion in an input signal, an amplitude value converter for converting an input signal value into another value by an Nth-order function conversion characteristic (N> 1),
A non-recursive filter that equalizes the output of the amplitude value converter, and the magnitude of the equalization error of the output of the non-recursive filter is used as a criterion, and the amplitude is minimized so that the equalization error is minimized. And a learning unit that learns and updates the tap coefficient of the acyclic filter and the coefficient that determines the characteristic of the value converter.

(2) In an equalizer for compensating for waveform distortion in an input signal, an amplitude value converter for converting the input signal value into another value by the Nth-order function conversion characteristic (N> 1),
A non-recursive filter that equalizes the output of the amplitude value converter, an equalization error calculator that calculates an equalization error of the output of the non-cyclic filter, and an equalization output from the equalization error calculator. An equalizer comprising: a coefficient for determining a characteristic of the amplitude value converter and a coefficient learning device for learning and updating a tap coefficient of the acyclic filter by using an error.

(3) A RAM table for converting the input signal value into another value in the equalizer for compensating the waveform distortion in the input signal, and the N-th order function conversion characteristic of the input signal value ( Amplitude value converter for converting to another value according to N> 1), and when determining the conversion characteristics of the RAM table,
An input switch that selects and outputs the output of the amplitude value converter, and otherwise selects and outputs the output of the RAM table, and a non-recursive filter that equalizes the output of the input switch. And an equalization error calculator that calculates the equalization error of the output of the non-recursive filter, and a coefficient that determines the characteristic of the amplitude value converter using the equalization error output from the equalization error calculator. , And the contents of the RAM table based on the coefficient learning device that learns and updates the tap coefficient of the acyclic filter and the coefficient that determines the characteristic of the amplitude value converter updated by the coefficient learning device. And a conversion pair generation unit for updating.

(4) In the above means (1) to (3), the amplitude value converter has a cubic function conversion characteristic (N =
3) is a feature.

(5) The magnetic recording / reproducing apparatus is characterized by including the equalizer of the above-mentioned means (1) to (4) and using a head using a magnetoresistive effect as a reproducing head.

[0030]

According to the above-mentioned means (1) to (4), in the equalizer for compensating the waveform distortion of the input signal, the signal waveform of the input signal by the amplitude value converter and its square to Nth power Are multiplied by a coefficient by a multiplier, added by an adder, and input to a non-recursive filter that performs waveform equalization such as PR (partial response) characteristics.

Further, in the learning section, the magnitude of the equalization error of the output of the non-recursive filter is used as a criterion, and a coefficient for determining the characteristic of the amplitude value converter so that the equalization error is minimized,
Also, the tap coefficient of the acyclic filter is learned.

Conventionally, it is general to learn the tap coefficient of the non-recursive filter which constitutes the waveform equalizer with reference to the minimum equalization error, and the amplitude converter according to the present invention is the same as this. Since it is possible to learn the coefficient of the amplitude value converter using an algorithm, it is easy to apply to the conventional signal processing method because the learning algorithm is common with the waveform equalizer with the conventional learning function. Is.

Thus, by using the equalizer of the present invention, it becomes possible to adaptively compensate for different waveform distortions in the input signal.

According to the above-mentioned means (5), the magnetic recording / reproducing apparatus using the head (MR head) using the magnetoresistive effect for the reproducing head is the equalizer of the means (1) to (4). To have.

Here, the amplitude value converter of the equalizer, when the magnetic field-reproduction output conversion characteristic of the MR head is non-linear,
The reproduction signal reproduced by the MR head is converted into a value based on the amplitude value converter characteristic which is the inverse characteristic of the magnetic field-reproduction output conversion characteristic of the MR head and is output.

Therefore, the combined characteristic of the magnetic field-reproduction output conversion characteristic of the MR head and the amplitude value converter characteristic of the amplitude value converter is substantially linear.

As a result, it becomes possible to adaptively compensate the waveform distortion in the reproduction signal due to the non-linearity of the magnetic field-reproduction output conversion characteristic of the MR head.

Further, it is possible to easily cope with MR heads having different nonlinear magnetic field-reproduction output conversion characteristics by learning without measuring the waveform distortion in the reproduction signal reproduced by the MR head. Become.

Further, even when the change with time of the MR head poses a problem, the coefficient can be updated by learning as needed.

[0040]

Embodiments of the present invention will be described below in detail with reference to the drawings.

In all the drawings for explaining the embodiments, parts having the same function are designated by the same reference numerals, and the repeated description thereof will be omitted.

[Embodiment 1] FIG. 1 is a block diagram showing a schematic configuration of a non-linear compensation equalizer which is an embodiment (Embodiment 1) of the present invention.

In FIG. 1, 1 is an amplitude value converter, 2 is F
IR filter (non-cyclic filter), 3 is an equalization error calculator, 4 coefficient update learning unit, 7 is an adder, and 10 is a multiplier.

The non-linear compensation equalizer of this embodiment is an equalizer for compensating the waveform distortion in the reproduced signal reproduced by the MR head, and the amplitude value converter 1 has an Nth-order function conversion characteristic (N>
1).

The amplitude value converter 1 multiplies the signal waveform 8 of the reproduced signal and the results from the square to the Nth power thereof by the coefficients (c2 to cN) by the multiplier 10, and adds them. Add by 7.

The output of the amplitude value converter 1 is input to the FIR filter 2, and the equalization error calculator 3 obtains the equalization error 5 which is the difference between the output 6 of the FIR filter 2 and the equalization target. .

The FIR filter 2 constitutes a waveform equalizer, and performs waveform equalization giving characteristics such as PR (partial response) characteristics.

Further, the coefficient update learning unit 4 uses the equalization error 5 and tap coefficients for determining the characteristics of the FIR filter and the coefficient (c) of the amplitude value converter so as to minimize the equalization error.
2 to cN) are calculated and updated.

FIG. 2 is a diagram for explaining the magnetic field-reproduction output conversion characteristic of the MR head, and FIG. 3 is a diagram for explaining the amplitude value conversion characteristic of the amplitude converter 1 shown in FIG. .

The amplitude value converter 1 shown in FIG. 1 converts the reproduction signal reproduced by the MR head into a value based on the amplitude value conversion characteristic and outputs the value.

As shown in FIG. 2, when the magnetic field-reproduction output conversion characteristic 84 showing the reproduction output with respect to the input magnetic field of the MR head deviates from the linear conversion characteristic 83, the amplitude value converter 1 shown in FIG. To perform the amplitude value converter 1
The amplitude value conversion characteristics of M are as shown in FIG.
The amplitude value conversion characteristic 93, which is the inverse characteristic of the magnetic field-reproduction output conversion characteristic 84 of the R head, is approached.

Therefore, the magnetic field-reproduction output conversion characteristic 84 of the MR head and the amplitude value converter characteristic 9 of the amplitude value converter 1
The characteristic obtained by combining 3 and 3 is almost linear.

As a result, it becomes possible to adaptively compensate the waveform distortion in the reproduction signal due to the non-linearity of the magnetic field-reproduction output conversion characteristic of the MR head.

Conventionally, it is general to learn the equalizer tap coefficient based on the minimum equalization error, and the amplitude converter 1 according to the present embodiment uses the same algorithm as the amplitude value. Since the coefficient of the converter 1 can be learned, it can be easily added to the conventional waveform equalizer composed of a non-recursive filter.

Further, it is possible to easily cope with each MR head having different nonlinear magnetic field-reproduction output conversion characteristics by learning without measuring the waveform distortion in the reproduction signal reproduced by the MR head. Becomes

Further, even when the change with time of the MR head poses a problem, the coefficient can be updated by learning as needed.

[Embodiment 2] FIG. 4 shows a schematic construction of a non-linear compensation equalizer which is another embodiment (Embodiment 2) of the present invention and a schematic construction of a magnetic disk device using the same. It is a block diagram.

In FIG. 4, reference numeral 21 denotes a magnetic recording medium, and 22
Is an MR head, 23 is a reproduction amplifier, 24 is a low-pass filter (LPF), 25 is an amplitude converter, 26 is a FIR filter, 27 is a Viterbi decoder, 28 is an equalization error calculator,
29 is an LMS (Least Mean Square) algorithm coefficient learning device, and 36 is an A / D converter (A /
D converter).

In the magnetic disk device shown in FIG. 4, the input magnetic field from the magnetic recording medium 21 is detected by the MR head 22, the reproduction signal reproduced by the MR head 22 is amplified by the reproduction amplifier 23, and the low pass filter is obtained. (LPF) 2
After passing through 4, it is quantized by an A / D converter (ADC) 36.

The reproduced signal quantized by the AD converter (ADC) 36 is input to the non-linear compensation equalizer 32.

Here, the non-linear compensation equalizer 32 includes an amplitude converter 25, a 9-tap FIR filter 26, an equalization error calculator 28, and an LMS (Least Mean Squ).
are) algorithm coefficient learning device 29.

In the non-linear compensation equalizer 32, the characteristics of the partial response class 4 detection method are targeted for equalization.

The partial response class 4 detection method is a method that can be described by the following expression (1) when the 1-bit delay operator is expressed by D.

[0064]

## EQU1 ## (1-D) * (1 + D) = 1-D ** 2 ...
(1) Here, the 1-bit delay operator D means the bit interval in the signal sequence of the recording / reproducing system, and D ** 2 means the square of D.

The output of the non-linear compensation equalizer 32 is input to the Viterbi decoder 27, the maximum likelihood decoding is performed in the Viterbi decoder 27, and the data series recorded in the magnetic recording medium 21 is detected.

Although not shown in FIG. 4, the output of the Viterbi decoder 27 is input to the recording decoder shown in FIG.

The LMS algorithm has an equalization error of 3
This is an algorithm that updates the tap coefficient so that the equalization error 30 is minimized by using 0 and the tap input value.

FIG. 5 is a block diagram of the non-linear compensation equalizer 32 shown in FIG.
FIG. 3 is a block diagram showing a detailed configuration of FIG.

In FIG. 5, 41 and 45 are coefficient values and 42
Is a multiplier, 43 is a delay element, 44 is an adder, 46 is an equalization target, and 50 is a coefficient learning circuit controller.

As shown in FIG. 5, the amplitude value converter 25 has 3
The signal waveform 35 of the reproduction signal reproduced by the MR head 22 having the following function conversion characteristic is multiplied by the power calculated by combining the multiplier 42 with the coefficient value 41 (c2, c3) by the coefficient unit 56, respectively. Multiply and add the result
Add with.

Generally, if the amplitude value converter 25 has a cubic function conversion characteristic, it should sufficiently compensate the waveform distortion in the reproduction signal due to the non-linearity of the magnetic field-reproduction output conversion characteristic of the MR head 22. Is possible.

FIR filter 26 forming a PR equalizer
Is composed of a delay element 43 and a coefficient unit 47, and performs waveform equalization giving a PR characteristic.

The difference between the output 31 of the FIR filter 26 and the equalization target 46 is obtained by the equalization error calculator 28, and the resulting equalization error 30 is input to the LMS algorithm coefficient learner 29.

In the LMS algorithm coefficient learning device 29,
The tap input value 48 input to each coefficient unit 47 of the FIR filter 26 is the FIR filter tap input value sequence 52 (h
1_in, ..., H9_in) is given to the coefficient learning circuit control unit 50.

The coefficient learning circuit controller 50 controls the equalization error 30
The FIR is adjusted in time with the time when is calculated.
The filter tap input value sequence 52 is output as the tap input value 51, and the tap input value 51 and the equalization error 30 are output.
And a step size parameter 52 for controlling the learning speed and stability for each tap.
(Myu) is multiplied, and the result is added to the immediately preceding coefficient value held using the delay element.

Based on this result, the coefficient learning circuit control unit 50
Generates an FIR filter coefficient update command 53 to update the coefficient value 45 (h1 to h9).

Similarly to the amplitude value converter 25, the tap input value 49 to each coefficient unit 56 is given to the coefficient learning circuit control unit 50 as the amplitude value converter tap input value sequence 54, and the like. The FIR filter 26 using the digitization error 30.
The coefficient is calculated in the same manner as, and the amplitude value converter coefficient update command 5
5 updates the coefficient value 41 of the amplitude value converter 25.

FIG. 6 is a block diagram showing an example of the equalization target 46 shown in FIG.

The equalization target 46 shown in FIG. 6 is composed of a ternary decision unit 71 to which the output 31 of the FIR filter 26 is input. The output 31 of the FIR filter 26 and the ternary decision unit 71 are included.
The difference from the output from is output as the equalization error 30.

FIG. 7 is a diagram showing the effect of the non-linear compensation equalizer of this embodiment.

In FIG. 7, 101 is an equalization error when equalization is performed by the conventional equalizer, and 102 is an equalization error when the nonlinear compensation equalizer of this embodiment is used. ,
Equalization error 1 when the nonlinear compensation equalizer of this embodiment is used
02 is an equalization error 101 when equalized by a conventional equalizer.
It is much smaller than.

In FIG. 7, the vertical asymmetry degree As
Is obtained by using the positive and negative peak values Ep and En of the reproduced waveform at low density where the waveform interference is sufficiently small.

[0083]

## EQU00002 ## As = (Ep-En) / (Ep + En) * 1
00 (%) ... It is a value defined by (2).

In this embodiment, the case where the non-linear compensation equalizer of the present invention is applied to the digital signal processing system using the A / D converter (ADC) 36 has been described.
The non-linear compensation equalizer of the present invention can also be applied to an analog signal processing system using a transversal filter as a PR equalizer giving a PR characteristic.

Further, the present invention does not limit the recording system of the magnetic recording medium 21 and can be applied to any recording system.

As described above, according to the second embodiment, it is possible to adaptively compensate the waveform distortion in the reproduction signal due to the non-linearity of the magnetic field-reproduction output conversion characteristic of the MR head 22.

Further, without measuring the waveform distortion in the reproduced signal reproduced by the MR head 22, the MR head 22 has different non-linear magnetic field-reproduction output conversion characteristics.
Each time, it becomes possible to easily cope by learning.

Further, even when the temporal change of the MR head 22 poses a problem, the coefficient can be updated by learning as needed.

As a result, the MR head 22 having a non-linear magnetic field-reproduction output conversion characteristic can be combined with various conventional signal processing systems, and the performance of the magnetic recording / reproducing apparatus can be improved.

On the other hand, it is possible to use the MR head 22 having a non-linear magnetic field-reproduction output conversion characteristic while maintaining the performance, and it is possible to improve the yield and reduce the manufacturing cost. It can be said that the economic effect is great.

[Embodiment 3] FIG. 8 is a block diagram showing a schematic configuration of a non-linear compensation equalizer which is another embodiment (Embodiment 3) of the present invention.

In FIG. 8, 61 is a RAM (Random).
m Access Memory table, 62 is an input switcher, 65 is an amplitude value conversion pair generation unit, 66 is an amplitude value converter, and 67 is a learning amplitude value conversion unit.

The non-linear compensation equalizer of the third embodiment is composed of a RAM table type non-linear compensation equalizer 68, and is used in the magnetic disk device as in the second embodiment.

In the RAM table type non-linear compensation equalizer 68 of this embodiment, during learning, the input switch 62 is switched to the learning amplitude converter 67 side, and the equalization error is minimized as in the case of the second embodiment. FIR filter 2
6 and the amplitude value converter 6 in the learning amplitude value converter 67
The coefficient of 6 is learned and updated.

In the learning amplitude value conversion unit 67, the input amplitude and output amplitude conversion pair is generated by the amplitude value conversion pair generation unit 6 after the learning is completed.
5, the table content update command 68 is generated accordingly, and the contents of the RAM table 61 are set.

In this case, it is also possible to record a fixed data series (reference pattern) in the magnetic recording medium 21 in advance and reproduce the fixed data series from the magnetic recording medium 21 during the learning. Is.

In this case, the equalization target 46 used in the equalization error calculator 28 may be composed of a memory in which the values of the constant data series output from the FIR filter 26 in sequence are stored. is there.

Except during learning, the input switch 62 is set to RA.
Switching to the M table 61 side, the amplitude value conversion is performed by the RAM table 61.

When the learning amplitude value converter 67 is used during learning and the RAM table 41 is used during data demodulation, a higher speed operation is possible and the data transfer rate is improved.

In the second and third embodiments, the PR in which the partial response class 4 system and the Viterbi decoder are combined is used.
The case where the non-linear compensation equalizer of the present invention is applied to the ML system has been described. As a system in which the PRML system is expanded, the pulse response up to the detector input is as shown in the following formulas (3) to (5). There is also a method.

[0101]

[Equation 3] (1-D) ((1 + D) ** 2) …………
……………… (3)

[0102]

[Equation 4] (1-D) ((1 + D) ** 3) …………
……………… (4)

[0103]

(5) (1-D) (1 + 2D + 2D ** 2 + D **
3) ... (5) It goes without saying that the non-linear compensation equalizer according to the present invention can be applied to the systems shown in the above equations (3) to (5).

As described above, the invention made by the present inventor is
Although the present invention has been specifically described based on the above-mentioned embodiments, the present invention is not limited to the above-mentioned embodiments, and it goes without saying that various modifications can be made without departing from the scope of the invention.

[0105]

The effects obtained by the typical ones of the inventions disclosed in the present application will be briefly described as follows.

(1) Since the equalizer of the present invention shares a learning algorithm with the conventional waveform equalizer with a learning function, it can be easily applied to a conventional signal processing system.

Further, by using the equalizer of the present invention, it is possible to adaptively compensate the waveform distortion due to different nonlinear characteristics in the input signal.

(2) According to the magnetic recording / reproducing apparatus of the present invention, it becomes possible to adaptively compensate the waveform distortion in the reproduced signal due to the non-linearity of the magnetic field-reproduction output conversion characteristic of the MR head. It is possible to easily cope with each MR head having different non-linear magnetic field-reproduction output conversion characteristics by measuring without measuring the waveform distortion in the reproduction signal reproduced by.

Further, even when the change with time of the MR head poses a problem, the coefficient can be updated by learning as needed.

(3) According to the magnetic recording / reproducing apparatus of the present invention, it is possible to combine an MR head having a non-linear magnetic field-reproducing output conversion characteristic with various conventional signal processing systems, and It is possible to improve performance.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration of a non-linear compensation equalizer that is an embodiment (Embodiment 1) of the present invention.

FIG. 2 is a diagram for explaining a magnetic field-reproduction output conversion characteristic of an MR head.

FIG. 3 is a diagram for explaining amplitude value conversion characteristics of the amplitude converter 1 shown in FIG.

FIG. 4 is a block diagram showing a schematic configuration of a non-linear compensation equalizer which is another embodiment (Embodiment 2) of the present invention and a schematic configuration of a magnetic disk device using the same.

5 is a block diagram showing a detailed configuration of the non-linear compensation equalizer shown in FIG.

FIG. 6 is a block diagram showing an example of the equalization target shown in FIG.

FIG. 7 is a diagram showing an effect of the non-linear compensation equalizer of the present embodiment.

FIG. 8 is a block diagram showing a schematic configuration of a non-linear compensation equalizer which is another embodiment (third embodiment) of the present invention.

FIG. 9 is a diagram showing a magnetic field-reproduction output conversion characteristic of the MR head.

FIG. 10 is a diagram for explaining an outline of PRML signal processing.

[Explanation of symbols]

1, 66 ... Amplitude value converter, 2 ... FIR filter (non-recursive filter), 3 ... Equalization error calculator, 4 ... Coefficient update learning unit, 7 ... Adder, 10 ... Multiplier, 21 ... Magnetic recording medium ,
22 ... MR head, 23 ... Reproducing amplifier, 24 ... Low-pass filter (LPF), 25 ... Amplitude value converter, 26 ... FI
R filter, 27 ... Viterbi decoder, 28 ... Equalization error calculator, 29 ... LMS algorithm coefficient learning device, 36 ... A /
D converter, 41, 45 ... Coefficient value, 42 ... Multiplier, 4
3 ... Delay element, 44 ... Adder, 46 ... Equalization target, 50 ...
Coefficient learning circuit control unit, 71 ... Three-value judgment unit, 61 ... RAM
Table, 62 ... Input switcher, 65 ... Amplitude value conversion pair generator, 67 ... Learning amplitude value converter, 601, Recording encoder, 602 ... Precoder, 603 ... Write compensation circuit, 604 ... Automatic gain control circuit ( AGC), 605 ... Low-pass filter, 606 ... A / D converter, 607 ... PR
(Partial response) Equalizer, 608 ... Viterbi decoding circuit, 609 ... Recording decoder.

Claims (5)

[Claims] 1. An equalizer for compensating waveform distortion in an input signal, wherein an input signal value is converted into an Nth-order function conversion characteristic (N>).
An amplitude value converter for converting the value to another value according to 1), and an acyclic filter for equalizing the waveform of the output of the amplitude value converter;
Using the magnitude of the equalization error of the output of the non-recursive filter as a criterion, a coefficient for determining the characteristic of the amplitude value converter so that the equalization error is minimized, and the non-recursive filter An equalizer having a learning unit that learns and updates tap coefficients. 2. An equalizer for compensating for waveform distortion in an input signal, wherein an input signal value is converted into an Nth-order function conversion characteristic (N>).
An amplitude value converter for converting the value to another value according to 1), and an acyclic filter for equalizing the waveform of the output of the amplitude value converter;
An equalization error calculator that calculates the equalization error of the output of the non-recursive filter, and a coefficient that determines the characteristic of the amplitude value converter using the equalization error output from the equalization error calculator, and , A coefficient learning device that learns and updates the tap coefficient of the non-cyclic filter. 3. An equalizer for compensating for waveform distortion in an input signal, a RAM table for converting an input signal value into another value, and an Nth-order function conversion characteristic (N > 1) Amplitude value converter for converting to another value, and when determining the conversion characteristics of the RAM table, the output of the amplitude value converter is selected and output; otherwise, the RAM table An input switch that selects and outputs an output, a non-recursive filter that equalizes the output of the input switch, and an equalization error calculator that calculates an equalization error of the output of the non-cyclic filter, A coefficient learning device that learns and updates the coefficient that determines the characteristic of the amplitude value converter and the tap coefficient of the non-recursive filter by using the equalization error output from the equalization error calculator; Updated with coefficient learner And a conversion pair generation unit for updating the contents of the RAM table based on the coefficient that determines the characteristic of the amplitude value converter. 4. The equalizer according to claim 1, wherein the amplitude value converter has a cubic function conversion characteristic (N = 3). 5. A magnetic recording / reproducing apparatus comprising the equalizer according to claim 1 and a head using a magnetoresistive effect as a reproducing head.