Disk drive margining read channel by biasing log-likelihood ratios of a nonbinary iterative decoder
View Patent ↗A disk drive is disclosed comprising a head actuated over a disk and control circuitry operable to sample a read signal emanating from the head to generate read samples. First vectors of log-likelihood ratios (LLRs) are generated in response to the read samples, wherein each vector corresponds to a nonbinary symbol q in a set of symbols. The first vectors of LLRs are biased to generate biased vectors of LLRs, and the biased vectors of LLRs are decoded into a data sequence, wherein the biased vectors of LLRs increase an error rate of the data sequence.
1. A disk drive comprising:
a disk;
a head actuated over the disk; and
control circuitry operable to:
sample a read signal emanating from the head to generate read samples;
generate first vectors of log-likelihood ratios (LLRs) in response to the read samples, wherein each vector corresponds to a nonbinary symbol q in a set of symbols;
bias the first vectors of LLRs to generate biased vectors of LLRs; and
decode the biased vectors of LLRs into a data sequence, wherein the biased vectors of LLRs increase an error rate of the data sequence.
2. The disk drive as recited in claim 1 , wherein the symbol q is in a set of symbols defined over a finite field GF(2 p ) where p is an integer greater than one.
3. The disk drive as recited in claim 1 , wherein a soft input/output detector generates the first vectors of LLRs.
4. The disk drive as recited in claim 1 , wherein the control circuitry is further operable to:
write a test pattern to the disk;
read the test pattern from the disk to generate the read signal; and
bias the first vectors of LLRs in response to the test pattern.
5. The disk drive as recited in claim 4 , wherein the control circuitry is further operable to bias the first vectors of LLRs according to:
LLR
n
b
(
q
)
=
{
0
q
=
0
log
α
1
-
p
(
q
)
1
-
α
p
(
q
)
+
LLR
n
(
q
)
q
≠
0
and
LLR
n
b
(
q
_
)
=
{
log
1
-
α
p
(
0
)
α
(
1
-
p
(
0
)
)
+
LLR
n
(
q
_
)
q
=
0
LLR
n
(
q
_
)
q
≠
0
where:
q represents a nonbinary symbol in the test pattern;
LLR n (q) represents one of the first vectors of LLRs corresponding to the nonbinary symbol q;
LLR n b (q) represents one of the biased vectors of LLRs corresponding to the nonbinary symbol q;
LLR n ( q ) represents one of the first vectors of LLRs corresponding to a nonbinary symbol not equal to q;
LLR n b ( q ) represents one of the biased vectors of LLRs corresponding to a nonbinary symbol not equal to q;
p(x) represents a probability that a nonbinary symbol x was received based on the read samples; and
α is a scalar between zero and one.
6. The disk drive as recited in claim 4 , wherein the control circuitry is further operable to bias the first vectors of LLRs according to:
LLR
n
b
(
q
)
=
{
0
q
=
0
log
α
1
-
p
c
1
-
α
p
c
+
LLR
(
q
)
q
≠
0
and
LLR
n
b
(
q
_
)
=
{
log
1
-
α
p
c
α
(
1
-
p
c
)
+
LLR
n
(
q
_
)
q
=
0
LLR
n
(
q
_
)
q
≠
0
where:
q represents a nonbinary symbol in the test pattern;
LLR n (q) represents one of the first vectors of LLRs corresponding to the nonbinary symbol q;
LLR n b (q) represents one of the biased vectors of LLRs corresponding to the nonbinary symbol q;
LLR n ( q ) represents one of the first vectors of LLRs corresponding to a nonbinary symbol not equal to q;
LLR n b ( q ) represents one of the biased vectors of LLRs corresponding to a nonbinary symbol not equal to q;
p c represents a probability that a correct nonbinary symbol was received; and
α is a scalar between zero and one.
7. The disk drive as recited in claim 1 , wherein an iterative decoder generates the first vectors of LLRs.
8. A method of operating a disk drive comprising a head actuated over a disk, the method comprising:
sampling a read signal emanating from the head to generate read samples;
generating first vectors of log-likelihood ratios (LLRs) in response to the read samples, wherein each vector corresponds to a nonbinary symbol q in a set of symbols;
biasing the first vectors of LLRs to generate biased vectors of LLRs; and
decoding the biased vectors of LLRs into a data sequence, wherein the biased vectors of LLRs increase an error rate of the data sequence.
9. The method as recited in claim 8 , wherein the symbol q is in a set of symbols defined over a finite field GF(2 p ) where p is an integer greater than one.
10. The method as recited in claim 8 , wherein a soft input/output detector generates the first vectors of LLRs.
11. The method as recited in claim 8 , further comprising:
writing a test pattern to the disk;
reading the test pattern from the disk to generate the read signal; and
biasing the first vectors of LLRs in response to the test pattern.
12. The method as recited in claim 11 , wherein the first vectors of LLRs are biased according to:
LLR
n
b
(
q
)
=
{
0
q
=
0
log
α
1
-
p
(
q
)
1
-
α
p
(
q
)
+
LLR
n
(
q
)
q
≠
0
and
LLR
n
b
(
q
_
)
=
{
log
1
-
α
p
(
0
)
α
(
1
-
p
(
0
)
)
+
LLR
n
(
q
_
)
q
=
0
LLR
n
(
q
_
)
q
≠
0
where:
q represents a nonbinary symbol in the test pattern;
LLR n (q) represents one of the first vectors of LLRs corresponding to the nonbinary symbol q;
LLR n b (q) represents one of the biased vectors of LLRs corresponding to the nonbinary symbol q;
LLR n ( q ) represents one of the first vectors of LLRs corresponding to a nonbinary symbol not equal to q;
LLR n b ( q ) represents one of the biased vectors of LLRs corresponding to a nonbinary symbol not equal to q;
p(x) represents a probability that a nonbinary symbol x was received based on the read samples; and
α is a scalar between zero and one.
13. The method as recited in claim 11 , wherein the first vectors of LLRs are biased according to:
LLR
n
b
(
q
)
=
{
0
q
=
0
log
α
1
-
p
c
1
-
α
p
c
+
LLR
(
q
)
q
≠
0
and
LLR
n
b
(
q
_
)
=
{
log
1
-
α
p
c
α
(
1
-
p
c
)
+
LLR
n
(
q
_
)
q
=
0
LLR
n
(
q
_
)
q
≠
0
where:
q represents a nonbinary symbol in the test pattern;
LLR n (q) represents one of the first vectors of LLRs corresponding to the nonbinary symbol q;
LLR n b (q) represents one of the biased vectors of LLRs corresponding to the nonbinary symbol q;
LLR n ( q ) represents one of the first vectors of LLRs corresponding to a nonbinary symbol not equal to q;
LLR n b ( q ) represents one of the biased vectors of LLRs corresponding to a nonbinary symbol not equal to q;
p c represents a probability that a correct nonbinary symbol was received; and
α is a scalar between zero and one.
14. The method as recited in claim 8 , wherein an iterative decoder generates the first vectors of LLRs.