US 7,173,783 B1 · Media noise optimized detector for magnetic recording

In a data channel for a data storage device having a magnetic data storage media and a transducer for recording/playback of data on the media, the channel including a detector having multiple states arranged for maximum likelihood decoding of a series of sample bits with pattern dependent noise and nonlinearities to form a series of detected bits, a method of determining branch metrics for branches of a trellis corresponding to the multiple states in the detector, wherein each branch j connects two of the states, the method comprising the steps of:

(a) determining an ideal value dj as a function of the corresponding sample bits;
(b) determining a difference between the ideal value dj and a sample value y represented by the corresponding sample bits;
(c) determining a branch metric scaling factor cj as a function of the noise variance for the corresponding sample bits;
(d) generating the branch metric BM as a function of the difference and the scaling factor cj;
and
(e) performing read-to-read adaptation of the ideal value dj using statistics accumulated for other dj values read from the media.

3 more elements

In a data channel for a magnetic recording and playback system, the channel including a detector having multiple states arranged for maximum likelihood decoding of a series of sample bits with pattern dependent noise and nonlinearities to form a series of detected bits, a method of determining branch metrics for branches of a trellis corresponding to the multiple states in the detector, wherein each branch j connects two of the states, the method comprising the steps of:

for each branch j:
(a) determining an ideal value dj as a function of the corresponding sample bits;
(b) determining a difference between the ideal value dj and a sample value y represented by the corresponding sample bits;
(c) determining a branch metric scaling factor cj as a function of the noise variance for the corresponding sample bits;
and
(d) generating the branch metric BM as a function of the difference and the scaling factor cj;
and
performing read-to-read adaptation of the ideal values dj's by:
accumulating statistics for the ideal values dj's while reading from a disk zone having multiple sectors;
and
for each ideal value dj, obtaining a new ideal value djnew as a function of the existing ideal value djold and an estimated ideal value djest wherein the estimated ideal value djest is the estimated value of the existing ideal value djold obtained while reading from the zone.

8 more elements

In a data channel for a magnetic recording and playback system, the channel including a detector having multiple states arranged for maximum likelihood decoding of a series of sample bits with pattern dependent noise and nonlinearities to form a series of detected bits, a method of determining branch metrics for branches of a trellis corresponding to the multiple states in the detector, wherein each branch j connects two of the states, the method comprising the steps of:

for each branch j:
(a) determining an ideal value dj as a function of the corresponding sample bits;
(b) determining a difference between the ideal value dj and a sample value y represented by the corresponding sample bits;
(c) determining a branch metric scaling factor cj as a function of the noise variance for the corresponding sample bits;
and
(d) generating the branch metric BM as a function of the difference and the scaling factor cj;
wherein the branch metric for each branch j is represented by the relations:
w=2 loga(y−dj)−loga(cj),
BM=aw
wherein a is the logarithm base.

7 more elements

In a data channel for a data storage device having a magnetic data storage media and a transducer for recording/playback of data on the media, the channel including a detector having multiple states arranged for maximum likelihood decoding of a series of sample bits with pattern dependent noise and nonlinearities to form a series of detected bits, a method of determining branch metrics for branches of a trellis corresponding to the multiple states in the detector, wherein each branch j connects two of the states, the method comprising the steps of:

(a) determining an ideal value dj as a function of the corresponding sample bits for each branch j;
(b) detecting if there is degradation in the recorded data due to thermal decay, and if so, adapting the ideal value dj as a function of the thermal decay;
(c) determining a difference between the ideal value dj and a sample value y represented by the corresponding sample bits;
(d) determining a branch metric scaling factor cj as a function of the noise variance for the corresponding sample bits;
and
(e) generating the branch metric BM as a function of the difference and the scaling factor cj.

3 more elements

In a data channel for a data storage device having a magnetic data storage media and a transducer for recording/playback of data on the media, the channel including a detector having multiple states arranged for maximum likelihood decoding of a series of sample bits with pattern dependent noise and nonlinearities to form a series of detected bits, a method of determining branch metric for branches of a trellis corresponding to the multiple states in the detector, wherein each branch j connects two of the states, the method comprising the steps of:

(a) determining an ideal value dj as a function of the corresponding sample bits for each branch j;
(b) detecting if there is degradation in the recorded data due to external magnetic fields, and if so, adapting the ideal value dj as a function of the corresponding sample bits affected by the external magnetic fields;
(c) determining a difference between the ideal value dj and a sample value y represented by the corresponding sample bits;
(d) determining a branch metric scaling factor cj as a function of the noise variance for the corresponding sample bits;
and
(e) generating the branch metric BM as a function of the difference and the scaling factor cj.

3 more elements

A detector circuit for detecting data from signal samples in a data channel, wherein the data is read from a recording media, the detector circuit comprising:

a detector having multiple states arranged for maximum likelihood decoding of a series of sample bits with pattern dependent noise and nonlinearities to form a series of detected bits;
a metric generator that determines branch metrics for branches of a trellis corresponding to the multiple states in the detector, wherein each branch j connects two of the states, and for each branch j the metric generator determines an ideal value dj as a function of the corresponding sample bits, determines a difference between the ideal value dj and a sample value y represented by the corresponding sample bits, determines a branch metric scaling factor cj as a function of the noise variance for the corresponding sample bits, and generates the branch metric BM as a function of the difference and the scaling factor cj;
and
an optimizer that performs read-to-read adaptation of the ideal values dj's by accumulating statistics for the ideal values dj's while reading from a disk zone having multiple sectors, and for each ideal value dj, obtaining a new ideal value djnew as a function of the existing ideal value djold and an estimated ideal value djest wherein the estimated ideal value djest is the estimated value of the existing ideal value djold obtained while reading from the zone.

1 more element

A detector circuit for detecting data from signal samples in a data channel, wherein the data is read from a recording media, the detector circuit comprising:

a detector having multiple states arranged for maximum likelihood decoding of a series of sample bits with pattern dependent noise and nonlinearities to form a series of detected bits;
and
a metric generator that determines branch metrics for branches of a trellis corresponding to the multiple states in the detector, wherein each branch j connects two of the states, and for each branch j the metric generator determines an ideal value dj as a function of the corresponding sample bits, determines a difference between the ideal value dj and a sample value y represented by the corresponding sample bits, determines a branch metric scaling factor cj as a function of the noise variance for the corresponding sample bits, and generates the branch metric BM as a function of the difference and the scaling factor cj;
wherein the branch metric for each branch j is represented by the relations:
w=2 loga(y−dj)−loga(cj),
BM=aw
wherein a is the logarithm base.

4 more elements

A detector circuit for detecting data from signal samples in a data channel, wherein the data is read from a recording media, the detector circuit comprising:

a detector having multiple states arranged for maximum likelihood decoding of a series of sample bits with pattern dependent noise and nonlinearities to form a series of detected bits;
and
a metric generator that determines branch metrics for branches of a trellis corresponding to the multiple states in the detector, wherein each branch j connects two of the states, and the metric generator determines an ideal value dj as a function of the corresponding sample bits, detects if there is degradation in the recorded data due to thermal decay, and if so, adapts the ideal values as a function of the thermal decay, determines a difference between the ideal value dj and a sample value y represented by the corresponding sample bits, determines a branch metric scaling factor cj as a function of the noise variance for the corresponding sample bits, and generates the branch metric BM as a function of the difference and the scaling factor cj.

A detector circuit for detecting data from signal samples in a data channel, wherein the data is read from a recording media, the detector circuit comprising:

a detector having multiple states arranged for maximum likelihood decoding of a series of sample bits with pattern dependent noise and nonlinearities to form a series of detected bits;
and
a metric generator that determines branch metrics for branches of a trellis corresponding to the multiple states in the detector, wherein each branch j connects two of the states, and the metric generator determines an ideal value dj as a function of the corresponding sample bits for each branch j, detects if there is degradation in the recorded data due to external magnetic fields, and if so, adapts the ideal values as a function of the corresponding sample bits affected by the external magnetic fields, determines a difference between the ideal value dj and a sample value y represented by the corresponding sample bits, determines a branch metric scaling factor cj as a function of the noise variance for the corresponding sample bits, and generates the branch metric BM as a function of the difference and the scaling factor cj.