Frequency Domain Approach for Efficient Computation of Fixed-point Equalization Targets

Abstract

Various embodiments of the present invention provide systems and methods for equalizing an input signal. For example, various embodiments of the present invention provide a method for performing equalization in a storage device. Such methods include providing an equalizer circuit that is governed by a target value, and a filter circuit that is governed by a filter coefficient. An initial value is provided to the equalizer circuit as the target value, and an overall target based at least in part on the initial value and the filter coefficient is calculated. An updated value is calculated based on the overall target, and the updated value is provided to the equalizer circuit as the target value.

Description

BACKGROUND OF THE INVENTION

The present inventions are related to systems and methods for equalizing an input data set.

Reading information from a magnetic storage medium includes receiving an analog electrical signal representing the received information by a read channel circuit that, among other things, equalizes the incoming signal to a primary target. As an example for channels with media noise, an equalization target generally consists of two parts: (1) a relatively short primary equalization target consisting of two or three taps, and (2) a bank of noise prediction filters including a number of individual noise prediction filters that are each tuned to a different data pattern. The overall equalization target consists of the convolution of the primary target with the noise prediction filters. Such a target is data-dependent.

In some cases, determining the appropriate overall target involves selecting what is expected to be a reasonable primary target. Once the primary target is selected using, for example, a mean-squared error (MSE) approach, the coefficients of the bank of noise prediction filters are adaptively selected. Such an approach provides reasonable performance, but does not result in error-rate optimizing solutions. To provide error-rate optimization, more complex and costly approaches may be employed. Such approaches typically rely on error rate based search algorithms to yield an optimum target. In particular, a primary target is established during a calibration phase prior to normal operation by individually selecting each of a number of possible primary targets and thereafter testing each of the selected primary targets. The respective error rate exhibited by each of the primary targets is stored, and the generated error rates are compared one against another to determine the optimum primary target. While such approaches are capable of yielding an optimized primary target, they can be too costly in terms of computational complexity and/or time for practical use.

Hence, for at least the aforementioned reasons, there exists a need in the art for advanced systems and methods for determining equalization targets.

BRIEF SUMMARY OF THE INVENTION

The present inventions are related to systems and methods for equalizing an input data set.

Some embodiments of the present invention provide methods for performing equalization in a storage device. Such methods include providing an equalizer circuit that is governed by a target value, and a filter circuit that is governed by a filter coefficient. An initial value is provided to the equalizer circuit as the target value, and an overall target based at least in part on the initial value and the filter coefficient is calculated. An updated value is calculated based on the overall target, and the updated value is provided to the equalizer circuit as the target value.

In some instances of the aforementioned embodiments, calculating the updated value based on the overall target includes performing the following calculation:

$g_{m,k}=\sum _{l=0}^{\min \left(k,\mathrm{Nq}\right)}q_lh_{m,k-l},\mathrm{for}k=0,1,\dots ,N_g,\mathrm{and}m=0,1,\dots ,n.$

In one or more instances of the aforementioned embodiments, calculating the updated value based on the overall target includes calculating frequency matrices based on the overall target. In some such instances, calculating the updated value based on the overall target further includes computing a floating-point target based on the frequency matrices and determining a fixed-point target based on the calculated target. In some cases, determining the fixed-point target based on the floating-point target includes minimizing a cost function. Calculating the updated value based on the overall target may further include scaling the floating-point target. As an example, the methods may further include providing an analog to digital converter that provides a series of digital samples to the equalizer, and the scaling process is operable to conform the floating-point target to a range of the analog to digital converter.

Other embodiments of the present invention provide storage devices that include a storage medium, a read/write head assembly disposed in relation to the storage medium, and a read channel circuit. The read channel circuit includes an equalization circuit relying on a frequency domain approach for calculating a primary target. The equalization circuit equalizes a data signal derived from the storage medium via the read/write head assembly.

In some instances of the aforementioned embodiments, the equalization circuit includes an equalization filter, a noise prediction filter circuit, a primary target calculation circuit, and a fixed-point selection circuit. Operation of the equalization filter is governed by a target value, and operation of the noise prediction filter circuit is governed by a noise prediction coefficient. The primary target calculation circuit calculates an overall target based at least in part on a first value of the target value and the noise prediction coefficient. The fixed-point selection circuit determines a fixed-point target based on the earlier calculated target.

Yet other embodiments of the present invention provide equalization circuits that include an equalization filter, a noise prediction filter circuit, an overall target calculation circuit and a fixed-point selection circuit. Operation of the equalization filter is governed by a target value, and operation of the noise prediction filter circuit is governed by a noise prediction coefficient. The overall target calculation circuit calculates an overall target based at least in part on a first value of the target value and the noise prediction coefficient. The primary target calculation circuit calculates a primary target based at least in part on a first value of the overall target and a frequency matrix. The fixed-point selection circuit determines a fixed-point target based on the earlier calculated target. In some instances of the aforementioned embodiments, the equalization filter is implemented as a digital finite impulse response filter.

This summary provides only a general outline of some embodiments of the invention. Many other objects, features, advantages and other embodiments of the invention will become more fully apparent from the following detailed description, the appended claims and the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

A further understanding of the various embodiments of the present invention may be realized by reference to the figures which are described in remaining portions of the specification. In the figures, like reference numerals are used throughout several figures to refer to similar components. In some instances, a sub-label consisting of a lower case letter is associated with a reference numeral to denote one of multiple similar components. When reference is made to a reference numeral without specification to an existing sub-label, it is intended to refer to all such multiple similar components.

FIG. 1 depicts a data processing system utilizing equalization target determination in accordance with various embodiments of the present invention;

FIG. 2 is a flow diagram showing a method in accordance with some embodiments of the present invention for equalization target determination;

FIG. 3 shows a storage system including a read channel utilizing equalization target determination in accordance with various embodiments of the present invention; and

FIG. 4 shows a communication system including a receiver utilizing equalization target determination in accordance with different embodiments of the present invention.

DETAILED DESCRIPTION OF THE INVENTION

The present inventions are related to systems and methods for equalizing an input data set.

Various embodiments of the present invention provide systems and methods for determining equalization targets for a data transfer system. Such embodiments of the present invention rely on a frequency based approach for selecting a primary target. The frequency-domain based approach may be derived by letting g=[g₀, g₁, . . . , g_Ng]^T, where g denotes an overall target including a primary target convolved with a respective noise prediction target. The primary target is denoted by p=[p₀, p_i, . . . , p_Np]^T. In this case, N_p<N_g. The primary target, p, is designed such that the mismatch between the frequency responses of p and g are minimized in the least-square (LS) sense. The following LS cost function may be used to evaluate the mismatch:

$J\left(p\right)=\sum _{f\in F}{G\left(f\right)-P\left(f\right)}^2,\mathrm{with}G\left(f\right)=\sum _{k=0}^{\mathrm{Ng}}g_k{}^{-\mathrm{j2\pi }\mathrm{fk}},\mathrm{and}P\left(f\right)=\sum _{k=0}^{Np}p_k{}^{-\mathrm{j2\pi }\mathrm{fk}},$

where G(f) and P(f) denote the frequency response of g and p, respectively, and F denotes a grid of normalized frequency spanning the range form 0.0 to 0.5.

The vectors e_g(f) and e_p(f) are defined as:

e _g(f)=[1, e^−j2πf, e^−j4πf, e^−j6πf, . . . , e^−jNgπf]^T; and

e _p(f)=[1, e^−j2πf, e^−j4πf, e^−j6πf, . . . , e^−jNgπf]^T.

Using the preceding vector equations, G(f) and P(f) can be expressed as:

G(f)=g^Te_g(f); and

P(f)=p^Te_p(f).

Through substitution, the cost function can be expressed as:

$\begin{array}{c}J\left(p\right)=\sum _{f\in F}{g^Te_g\left(f\right)-p^Te_p\left(f\right)}^2\\ =g^TE_gg+p^TE_pp-g^TE_{\mathrm{pg}}^Hp-p^TE_{\mathrm{pg}}g,\end{array}$

where

$E_g=\sum _{f\in F}e_g\left(f\right){}_g^H\left(f\right),E_p=\sum _{f\in F}e_p\left(f\right){}_p^H\left(f\right),\mathrm{and}E_{\mathrm{pg}}=\sum _{f\in F}e_p\left(f\right){}_g^H\left(f\right).$

In these equations, the superscript ^H denotes a conjugate transpose, unlike the standard transpose ^T. In this equation, the target coefficients are presumed to be real numbers.

The optimum floating-point primary target that minimizes the cost function J(p) can be determined by setting the gradient of J(p) with respect to p to zero and solving for p. Doing this, the following optimum floating-point target is defined by the following equation:

p _opt=(E_p+E_p*)⁻¹(E_pg+E_pg*)g,   Equation (1)

where the superscript * denotes a complex conjugation.

The matrices E_p and E_pg can be computed as follows. First, let the frequency grid be: F={0, Δ, 2Δ, 3Δ, . . . , LΔ}, where Δ=0.5/L for a reasonably large positive integer L. Next, the quantity E(k,l) is defined as:

$\begin{array}{cc}\begin{array}{c}E\left(k,l\right)=\sum _{f\in F}{}^{-\mathrm{j2\pi }\mathrm{fk}}{}^{\mathrm{j2\pi }\mathrm{fl}}\\ =\sum _{m=0}^L{}^{\mathrm{j2\pi }m\left(l-k\right)\Delta }\\ =\frac{1-{}^{\mathrm{j2\pi \Delta }\left(L+1\right)}}{1-{}^{\mathrm{j2\pi \Delta }}},\mathrm{for}k\ne l,\mathrm{and}\\ =L+1,\mathrm{for}k=l.\end{array}& \mathrm{Equation}\left(2\right)\end{array}$

Then, the (k,l)^th elements of the matrices (E_p+E_p*) and (E_pg+E_pg*) can be obtained for k=0, 1, . . . , N_p as:

(E_p+E_p*)_(k,l)=E(k,l)+E*(k,l) for l=0, 1, . . . , N_p; and

(E_pg+E_pg*)_(k,l)=E(k,l)+E*(k,l) for l=0, 1, . . . , N_g.

Using equation (2) above, the following may be derived:

$\begin{array}{c}E\left(k,l\right)+E^{*}\left(k,l\right)=\frac{1-\cos \left(\pi \left(l-k\right)/L\right)-\cos \left(\pi \left(L+1\right)/L\right)+{\left(-1\right)}^{l-k}}{1-\cos \left(\pi \left(l-k\right)/L\right)}\\ =1+{\left(-1\right)}^{l-k}\mathrm{for}k\ne l;\mathrm{and}\\ =2L+2\mathrm{for}k=l.\end{array}$

Of note, if L is allowed to trend toward infinity, then (E_p+E_p*) becomes the (N_p+1)×(N_p+1) identity matrix I_Np+1, and (E_pg+E_pg*) takes the form [I_Np+1 O_Ng−Np] where O_Ng−Np is a (N_p+1)×(N_g−N_p) matrix of zeros. This would make the optimum floating-point primary target P_opt equal to the first N_p+1 coefficients of the given floating-point target g. Thus, by choosing L to be finite (e.g., between one hundred (100) and two hundred (200)), we are in effect introducing a small perturbation in the primary target compared to g.

The target coefficients that result from equation (1) are scaled to satisfy the range of requirements of an upstream analog to digital converter. This is done by scaling the coefficients as set forth below:

$p_{\mathrm{opt},k}\leftarrow p_{\mathrm{opt},k}*\mathrm{Round}\left(\frac{S_{\mathrm{adc}}}{\sum _{l=0}^{Np}p_{\mathrm{opt},l}}\right),$

where S_adc is the maximum signal amplitude expected at the output of the equalizer and p_opt,k is the k^th tap of the target vector p_opt, for k=0, 1, . . . , N_p. The aforementioned scaling operation may not be necessary if the given target, g, is already of the appropriate scale.

The target coefficients that result from the scaling process need to be converted into fixed-point values. Fixed-point values are obtained by selecting a fixed-point target that is closest to p_opt. This is achieved by minimizing the following least-square cost function:

$\tilde{J}\left(\tilde{p}\right)=\sum _{k=0}^{Np}{\left(\frac{p_{\mathrm{opt},k}}{\sum _{l=0}^{Np}p_{\mathrm{opt},l}}-\frac{{\tilde{p}}_k}{\sum _{l=0}^{Np}{\tilde{p}}_l}\right)}^2.$

The optimum fixed-point target coefficients that minimize the cost function {tilde over (J)}({tilde over (p)}) can be obtained by evaluating {tilde over (J)}({tilde over (p)}) over a grid around p_opt. Since p_opt is the floating-point optimum target and because the value of N_p is small, the computation cost and time required to determine the minimum of {tilde over (J)}({tilde over (p)}) is small. In some embodiments of the present invention, the value of N_p is one or two.

Turning to FIG. 1, a data processing system 100 utilizing equalization target determination is shown in accordance with various embodiments of the present invention. Data processing system 100 includes an analog to digital converter 110 that receives an analog input signal 110. Analog to digital converter 110 provides a stream of digital samples 115 corresponding to analog input 105. Digital samples 115 are provided to an equalization filter 120 that may be implemented as a digital finite infinite response filter. Equalization filter 120 is governed by a primary target 122. Depending upon the design of equalization filter 120, primary target 122 includes a defined number of filter taps. In one particular embodiment of the present invention, primary target 122 includes two filter taps. In another particular embodiment of the present invention, primary target 122 includes three filter taps. Based upon the disclosure provided herein, one of ordinary skill in the art will recognize different numbers of filter taps that may be included in primary target 122 depending upon the design of equalization filter 120.

An output 125 from equalization filter 120 is provided to a noise prediction filter block 130 and to an adaptive noise prediction coefficient calculation circuit 140. Noise prediction filter block 130 includes a number ‘n’ of noise prediction filters. The operation of noise prediction filter block 120 is governed by noise prediction coefficients 145. Each of the noise prediction filters provides a respective filtered output (i.e., NP0, NP1, . . . , NPn). In one particular embodiment of the present invention, noise prediction filter block 130 includes eight noise prediction filters. Based upon the disclosure provided herein, one of ordinary skill in the art will recognize different numbers of noise prediction filters that may be utilized in accordance with different embodiments of the present invention. The outputs from noise prediction filter block 130 (i.e., NP0, NP1, . . . , NPn) are provided to a detector 135, and to an adaptive noise prediction coefficient calculation circuit 140. Detector 135 may be any detector known in the art that is capable of receiving a raw input and performing a detection algorithm on the raw input in an attempt to generate an output 150 that reflects an originally transferred or stored data set. Adaptive noise prediction coefficient calculation circuit 140 may be any circuit known in the art that is capable of adaptively determining appropriate noise prediction coefficients based on output 150, equalizer output 125 and the outputs from the respective noise prediction filters (i.e., NP0, NP1, . . . , NPn).

An initial primary target 155 is selected as primary target 122 by switching an initial selection control 160 to a multiplexer 170. Initial primary target 155 and noise prediction coefficients 145 are provided to a primary target calculation circuit 175. Initial primary target 155 may be a set of static filter taps that would be expected to provide reasonable performance by equalization filter 120. Noise prediction coefficients 145 are adaptively established as is known in the art some number of clock cycles after initial primary target 155 is applied to equalization circuit 120 via multiplexer 170. Primary target calculation circuit 175 calculates optimum primary target in floating-point format. The resulting target is provided to a scaling circuit 185 that scales the target values to match the scale of analog to digital converter 110. The resulting scaled floating-point optimum target is then used by an optimum fixed-point selection circuit 190 to determine fixed-point targets that are close to the optimum floating-point target in the least-square sense. A ranking circuit 195 selects a particular one of the fixed-point target values to provide as an updated primary target 197. Once updated primary target 197 is calculated, initial selection control 160 is switched such that updated primary target 197 is provided as primary target 122 via multiplexer 170.

In operation, initial selection control 160 is initially asserted to select initial primary target 155. Initial primary target 155 is provided to equalization filter 120 as primary target 122. Over a number of clock cycles, noise prediction coefficients 145 provided to noise prediction filter block 130 are adaptively calculated by adaptive noise prediction coefficient calculation circuit 140 using output 150, equalizer output 125 and the outputs from the respective noise prediction filters (i.e., NP0, NP1, . . . , NPn) using adaptive approaches known in the art. The noise prediction coefficients that are obtained are referred to as h_k, and are defined by the following vector:

h _k =[h _k,0 , h _k,1 , . . . , h _k,Nh]^T for k=0, 1, 2, . . . , n.

Noise prediction coefficients 145 and initial primary target 155 are provided to primary target calculation circuit 175 where an overall target is determined for each of the N_p filters in accordance with the following equation:

$\begin{array}{cc}{g}_{m,k}=\sum _{l=0}^{\min \left(k,\mathrm{Nq}\right)}q_lh_{m,k-l},\mathrm{for}k=0,1,\dots ,N_g,\mathrm{and}m=0,1,\dots ,n.& \mathrm{Equation}3\end{array}$

In this case, m denotes a particular filter. Using a frequency grid resolution of Δ=0.5/L where L is a large positive integer, the (k,l)^th elements of the (N_p+1)×(N_p+1) matrix, {tilde over (E)}_p, and the (N_p+1)×(N_g+1) matrix, {tilde over (E)}_pg are computed as:

1+(−1)^l−k for k≠l; and

2L+2 for k=l.

Of interest, due to the symmetry present in the matrices, {tilde over (E)}_p and {tilde over (E)}_pg, only the longest row or column of the matrices need be computed.

The optimum floating-point primary target is then calculated in accordance with the following equation:

p _opt ={tilde over (E)} _p ⁻¹ {tilde over (E)} _pg g,

where g=[g₀, g₁, . . . , g_Ng]^T is obtained using equation 3 above. The calculated target is then scaled by scaling circuit 185 in accordance with the following equation:

$p_{\mathrm{opt},k}\leftarrow p_{\mathrm{opt},k}*\mathrm{Round}\left(\frac{S_{\mathrm{adc}}}{\sum _{l=0}^{Np}p_{\mathrm{opt},l}}\right),$

where S_adc is the maximum signal amplitude expected at the output of the equalizer and p_popt,k is the k^th tap of the target vector p_opt, for k=0, 1, . . . , N_p. Next, the optimum fixed-point primary target is determined by optimum fixed-point selection circuit 190 to minimize the following cost function:

$\tilde{J}\left(\tilde{p}\right)=\sum _{k=0}^{Np}{\left(\frac{p_{\mathrm{opt},k}}{\sum _{l=0}^{Np}p_{\mathrm{opt},l}}-\frac{{\tilde{p}}_k}{\sum _{l=0}^{Np}{\tilde{p}}_l}\right)}^2,$

where {tilde over (p)}=[{tilde over (p)}₀, {tilde over (p)}₁, . . . , {tilde over (p)}_Np]^T corresponds to a point on the grid placed around the optimum floating-point target popt obtained from scaling circuit 185. The grid for each coefficient is given by:

{{tilde over (p)}_opt,k−M, {tilde over (p)}_opt,k−M+1, . . . , {tilde over (p)}_opt,k+M−1, {tilde over (p)}_opt,k+M}

for k=0, 1, . . . , N_p where {tilde over (p)}_opt,k=Round({tilde over (p)}_opt,k) and the integer M denotes the grid-size. The calculated optimum targets are then ordered in the order of increasing values of the cost function {tilde over (J)}({tilde over (p)}) by ranking circuit 195. By so ordering, several ranked choices for the primary target are available around Popt. One of the ranked choices is selected and provided as updated primary target 197.

Once updated primary target 197 is available, initial selection control 160 is switched causing multiplexer 170 to provide updated primary target 197 as primary target 122 to equalization filter 120. Some number of clock cycles after updated primary target 197 is applied to equalization filter 120, noise prediction coefficients 145 are adaptively updated as is known in the art.

Turning to FIG. 2, a flow diagram 200 shows a method in accordance with some embodiments of the present invention for equalization target determination. Following flow diagram 200, an initial primary target is selected to be provided as the primary target to an equalizer (block 205). The initial primary target may be a set of static filter taps that would be expected to provide reasonable performance by an equalization filter to which they are provided. Over a number of clock cycles, noise prediction coefficients are adaptively calculated to adjust to the output of the equalizer (block 210). This approach to adaptive calculation of noise prediction coefficients is known in the art. The noise prediction coefficients that are obtained are referred to as h_k, and are defined by the following vector:

h _k −[h _k0 , h _k,1 , . . . , h _k,Nh]^T for k=0, 1, 2, . . . , n.

The overall target is then calculated based on the noise prediction coefficients and the initial primary target (block 215). The overall target is determined for each of the noise prediction filters (i.e., N_p filters) in accordance with the following equation:

$g_{m,k}=\sum _{l=0}^{\min \left(k,\mathrm{Nq}\right)}q_lh_{m,k-l},\mathrm{for}k=0,1,\dots ,N_g,\mathrm{and}m=0,1,\dots ,n.$

In this case, m denotes a particular noise prediction filter. Using this information, frequency matrices are calculated (block 220). For example, using a frequency grid resolution of Δ=0.5/L where L is a large positive integer, the (k,l)^th elements of the (N_p+1)×(N_p+1) matrix, {tilde over (E)}_p, and the (N_p+1)×(N_g+1) matrix, {tilde over (E)}_pg are computed as:

1+(−1)^l−k for k≠l; and

2L+2 for k=l.

Of interest, due to the symmetry present in the matrices, {tilde over (E)}_p and {tilde over (E)}_pg, only the longest row or column of the matrices need be computed.

The optimum floating-point primary target is then calculated (block 225). This calculation may be done in accordance with the following equation:

p _opt ={tilde over (E)} _p ⁻¹ {tilde over (E)} _pg g,

where g=[g₀, g₁, . . . , g_Ng]^T is obtained using equation 3 above. In some cases, the calculated targets are then scaled to match an input analog to digital converter. This scaling may be performed in accordance with the following equation:

$p_{\mathrm{opt},k}\leftarrow p_{\mathrm{opt},k}*\mathrm{Round}\left(\frac{S_{\mathrm{adc}}}{\sum _{l=0}^{Np}p_{\mathrm{opt},l}}\right),$

where S_adc is the maximum signal amplitude expected at the output of the equalizer and p_opt,k is the k^th tap of the target vector p_opt, for k=0, 1, . . . , N_p. Next, the optimum fixed-point primary target is calculated (block 230). This may be done by minimizing the following cost function:

$\tilde{J}\left(\tilde{p}\right)=\sum _{k=0}^{Np}{\left(\frac{p_{\mathrm{opt},k}}{\sum _{l=0}^{Np}p_{\mathrm{opt},l}}-\frac{{\tilde{p}}_k}{\sum _{l=0}^{Np}{\tilde{p}}_l}\right)}^2,$

where {tilde over (p)}=[{tilde over (p)}₀, {tilde over (p)}₁, . . . , {tilde over (p)}_Np]^T corresponds to a point on the grid placed around the optimum floating-point target popt obtained from scaling circuit 185. The grid for each coefficient is given by:

{{tilde over (p)}_opt,k, −M, {tilde over (p)}_opt,k−M+1, . . . , {tilde over (p)}_opt,k+M−1, {tilde over (p)}_opt,k+M}

for k =0, 1, . . . , N_p where {tilde over (p)}_opt,k=Round({tilde over (p)}_opt,k) and the integer M denotes the grid-size. The calculated optimum targets are then ordered in the order of increasing values of the cost function {tilde over (J)}({tilde over (p)}) (block 235). By so ordering, several ranked choices for the primary target are available around p_opt. A selected one of the ranked choices may then be provided as the primary target to the equalizer and the noise prediction coefficients can be adaptively calculated to conform to the updated primary target.

Turning to FIG. 3, a storage system 400 including a read channel including frequency-domain approach for equalization target determination 410 is shown in accordance with various embodiments of the present invention. Storage system 400 may be, for example, a hard disk drive. The incorporated frequency domain approach for equalization target determination is capable of calculating an optimized primary target based on the frequency domain approach discussed above in relation to FIG. 2. In some embodiments, read channel module 410 may be implemented to include the circuitry discussed above in relation to FIG. 1. In addition, storage system 400 includes an interface controller 420, a preamplifier 412, a hard disk controller 466, a motor controller 468, a spindle motor 472, a disk platter 478, and a read/write head 476. Interface controller 420 controls addressing and timing of data to/from disk platter 478. The data on disk platter 478 consists of groups of magnetic signals that may be detected by read/write head assembly 476 when the assembly is properly positioned over disk platter 478. In a typical read operation, read/write head assembly 476 is accurately positioned by motor controller 468 over a desired data track on disk platter 478. Motor controller 468 both positions read/write head assembly 476 in relation to disk platter 478 and drives spindle motor 472 by moving read/write head assembly to the proper data track on disk platter 478 under the direction of hard disk controller 466. Spindle motor 472 spins disk platter 478 at a determined spin rate (RPMs).

Once read/write head assembly 476 is positioned adjacent the proper data track, magnetic signals representing data on disk platter 478 are sensed by read/write head assembly 476 as disk platter 478 is rotated by spindle motor 472. The sensed magnetic signals are provided as a continuous, minute analog signal representative of the magnetic data on disk platter 478. This minute analog signal is transferred from read/write head assembly 476 to read channel module 410 via preamplifier 412. Preamplifier 412 is operable to amplify the minute analog signals accessed from disk platter 478. In addition, preamplifier 412 is operable to condition the data from read channel module 410 that is destined to be written to disk platter 478. In turn, read channel module 410 decodes and digitizes the received analog signal to recreate the information originally written to disk platter 478. This data is provided as read data 403 to a receiving circuit. Prior to normal operation, read channel module 410 performs a frequency-based computation to determine equalization targets. A write operation is substantially the opposite of the preceding read operation with write data 401 being provided to read channel module 410. This data is then encoded and written to disk platter 478.

Turning to FIG. 4, a communication system 591 comprising a receiver 595 including frequency-domain approach for equalization target determination is depicted in accordance with different embodiments of the present invention. Communication system 591 includes a transmitter 593 that is operable to transmit encoded information via a transfer medium 597 as is known in the art. The encoded data is received from transfer medium 597 by receiver 595. Receiver 595 incorporates an equalization target calculation circuit similar to that described above in relation to FIGS. 1-2. Receiver 595 decodes and digitizes the received analog signal to recreate the information originally transmitted by transmitter 593. Prior to normal operation, receiver module 595 performs a frequency-domain based computation to determine equalization targets. Based on the disclosure provided herein, one of ordinary skill in the art will recognize a variety of mediums over which data transfer may be done, requiring a receiver with equalizer and target to recover the original data stream.

In conclusion, the invention provides novel systems, devices, methods and arrangements for determining equalization targets. While detailed descriptions of one or more embodiments of the invention have been given above, various alternatives, modifications, and equivalents will be apparent to those skilled in the art without varying from the spirit of the invention. Therefore, the above description should not be taken as limiting the scope of the invention, which is defined by the appended claims.

Claims

1. A method for performing equalization in a storage device, the method comprising: providing an equalizer circuit, wherein operation of the equalizer circuit is governed by a target value;providing a filter circuit, wherein the filter circuit is governed by a filter coefficient;providing an initial value to the equalizer circuit as the target value;calculating an overall target based at least in part on the initial value and the filter coefficient;calculating an updated value based on the overall target; andproviding the updated value to the equalizer circuit as the target value.

2. The method of claim 1, wherein calculating the updated value based on the overall target includes performing the following calculation:

3. The method of claim 1, wherein calculating the updated value based on the overall target includes: calculating frequency matrices based on the overall target.

4. The method of claim 3, wherein calculating the updated value based on the overall target further includes: computing a floating-point target based on the frequency matrices.

5. The method of claim 4, wherein calculating the updated value based on the overall target further includes: determining a fixed-point target based on the floating-point target.

6. The method of claim 5, wherein determining the fixed-point target based on the floating point target includes minimizing the following cost function:

7. The method of claim 4, wherein calculating the updated value based on the overall target further includes: scaling the floating-point target.

8. The method of claim 7, wherein the method further comprises: providing an analog to digital converter, wherein the analog to digital converter provides a series of digital samples to the equalizer; andwherein scaling the floating point target is operable to conform the floating-point target to a range of the analog to digital converter.

9. A storage device, the storage device comprising: a storage medium;a read/write head assembly disposed in relation to the storage medium; anda read channel circuit including an equalization circuit relying on a frequency domain approach for calculating a primary target; and wherein the equalization circuit equalizes a data signal derived from the storage medium via the read/write head assembly.

10. The storage device of claim 9, wherein the equalization circuit includes: an equalization filter, wherein operation of the equalization filter is governed by a target value;and a noise prediction filter circuit, wherein the noise prediction filter circuit is governed by a noise prediction coefficient;an overall target calculation circuit, wherein the overall target calculation circuit calculates an overall target based at least in part on a first value of the target value and the noise prediction coefficient;a primary target calculation circuit, wherein the primary target calculation circuit calculates a primary target based at least in part on a first value of the overall target and a frequency matrix; anda fixed-point selection circuit, wherein the fixed-point selection circuit determines a fixed-point target based on the floating-point target.

11. The storage device of claim 10, wherein calculating the overall target is done in accordance with the following equation:

12. The storage device of claim 10, wherein calculating the floating-point target based on the overall target is done in accordance with the following equation: {tilde over (p)}opt={tilde over (E)}p−1{tilde over (E)}pgg.

13. The storage device of claim 10, wherein determining the fixed-point target based on the floating-point target is done by minimizing the following cost function:

14. An equalization circuit, the circuit comprising: an equalization filter, wherein operation of the equalization filter is governed by a target value;and a noise prediction filter circuit, wherein the noise prediction filter circuit is governed by a noise prediction coefficient;an overall target calculation circuit, wherein the overall target calculation circuit calculates an overall target based at least in part on a first value of the target value and the noise prediction coefficient;a primary target calculation circuit, wherein the primary target calculation circuit calculates a primary target based at least in part on a first value of the overall target and a frequency matrix; anda fixed-point selection circuit, wherein the fixed-point selection circuit determines a fixed-point target based on the floating-point target.

15. The equalization circuit of claim 14, wherein calculating the overall target is done in accordance with the following equation:

16. The equalization circuit of claim 14, wherein calculating the floating-point target based on the overall target is done in accordance with the following equation: popt={tilde over (E)}p−1{tilde over (E)}pgg.

17. The equalization circuit of claim 14, wherein determining the fixed-point target based on the floating-point target is done by minimizing the following cost function:

18. The equalization circuit of claim 14, wherein the primary target calculation circuit receives an input from an analog to digital converter, and wherein the equalization circuit further comprises: a scaling circuit, wherein the scaling circuit is operable to scale the floating-point target to a range consistent with the analog to digital converter.

19. The equalization circuit of claim 18, wherein scaling the floating-point target to a range consistent with the analog to digital converter is done in accordance with the following equation:

20. The equalization circuit of claim 14, wherein the equalization filter is implemented as a digital finite impulse response filter.