Error propagation detection and control method, and memory...

Details Error propagation detection and control method, and memory... Error propagation detection and control method, and memory...

1999-11-18

2004-04-06

Pham, Chi (Department: 2631)

Pulse or digital communications

Equalizers

Automatic

C375S350000, C333S018000, C714S750000

Reexamination Certificate

active

06717986

ABSTRACT:

BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a method for the detection and control of error propagation based on multilevel decision feedback equalization (MDFE), and to a memory device that makes use of this method.
More particularly, it relates to a method for controlling error propagation by means of the results of error propagation detection and error propagation offset detection involving the detection of the offset direction of a signal due to error propagation.
2. Description of the Related Art
FIG. 15
is a block diagram of the structure of a magnetic disk device that makes use of multilevel decision feedback equalization (MDFE), and of the signal processing system in particular. In
FIG. 15
, an input NRZ signal composed of 1 and 0 bit strings is converted by a (
1
-
7
) encoder into a (
1
-
7
) code having a code rule in which the maximum continuance of same-polarity codes is 8 and the number of continuous alternations is 1.
The output of the (
1
-
7
) encoder has a value of (plus/minus)1 sampled at a timing of 1/T. With respect to the output of the (
1
-
7
) encoder, (1/1-D)mod2 is computed by the write FF circuit
2
, sent to the write/read head
3
, and written to a disk.
In
FIG. 15
, the data written to the disk is read by a write/read head
3
, and is amplified to a specific level through a head pre-amplifier circuit
4
and an AGC amplifier
5
.
The output of the AGC amplifier
5
is inputted to a forward filter
6
of an MDFE circuit
10
and converted into a ternary signal (−2, 0, +1). The output of a feedback filter
7
varies between input pulse polarity and reverse polarity. Specifically, the feedback filter
7
is assumed to be such that the polarity of the input reproduction signal pulses alternates.
Therefore, the output polarity of the feedback filter
7
is usually in the opposite direction from the polarity of the expected input to the forward filter
6
. An addition circuit
8
finds the difference between the output of the forward filter
6
and the output of the feedback filter
7
. As a result, the output of the addition circuit
8
has a waveform centered around a “0” level.
The output of the addition circuit
8
is then subjected to binary decision by a detector
9
. The output of the detector
9
is a binary coded sequence, and is put in a quaternary state by 1-bit convolution by the MDFE circuit
10
.
A (1-D)mod2 circuit
11
performs computation for write FF circuit
2
processing and inverse processing (1-D)mod 2, and decodes the (
1
-
7
) code by means of a (
1
-
7
) decoder
12
. This results in the reproduction of the read signal.
Thus, the structure in decision feedback equalization is such that the decision results from the detector
9
are fed back to the input side through the feedback filter
7
. This poses a problem in that an erroneous decision results in the propagation of the error.
One possible way to deal with this problem, as previously proposed by the inventors (PCT/98JP/05278), is to use a code conversion rule (d, k) to detect the occurrence of error propagation in decision feedback equalization, and make the slice level of a comparative decider
9
variable, so that error propagation is minimized.
FIGS. 16A and 16B
are diagrams illustrating an example of when the polarity remains the same for nine or more continuous symbols.
FIG. 16A
is a write data string. In contrast,
FIG. 16B
is a signal string that has been read, and is a case in which the signal is missing at the 100
th
sample. When a signal is missing, the output of the detector
9
remains fixed at a constant level.
In the example in
FIGS. 16A and 16B
, the error propagates from the point of the missing signal at the 100
th
sample, and the output of the detector
9
continues fixed at the same polarity of −1.
Meanwhile,
FIGS. 17A and 17B
are diagrams illustrating an example of continuous polarity alternations for two or more symbols.
FIG. 17A
is a write data string. In contrast,
FIG. 17B
is a signal string that has been read, and is a case in which the signal is missing at the 100
th
sample.
In the example in
FIG. 17
, the error propagates from the point of the missing signal at the 100
th
sample, and the output of the detector
9
is such that ten samples are fixed at +1 (interval I in FIG.
17
B), and this is followed by a repeating cycle of polarity alternations of two or more continuous symbols (interval II in FIG.
17
B).
The error in the read code strings illustrated in
FIGS. 16 and 17
is a code pattern that violates the code conversion rule (d, k). The propagation of error caused by this d constraint violation and k constraint violation can be suppressed by the method disclosed in the previous application (PCT/98JP/05278).
However, further study by the inventors has revealed that error propagation which occurs when the code conversion rule is satisfied cannot be detected and suppressed by the method proposed in the above-mentioned previous application.
An example of this is illustrated in
FIGS. 18A through 18C
.
FIG. 18A
illustrates the comparative decider input of the detector (solid line) and the decider slice level (dotted line).
FIG. 18B
illustrates the direct current offset error level.
FIG. 18C
illustrates the AGC error signal level.
In this example, when there is no error propagation, the pattern is a continuous “110000,” but when error propagation occurs, it is found to be a continuous pattern of “111001.”
SUMMARY OF THE INVENTION
An object of the present invention is therefore to provide an error propagation detection method and apparatus for use in decision feedback equalization type detection, with which it is possible to detect error propagation even when a specific code conversion rule is satisfied in an MDFE system.
The constitution of the present invention for achieving the stated object involves multilevel decision feedback equalization in which the output of a comparative decider for deciding the level of an input signal is fed back to the input side through a feedback filter using a slice level as a reference, and the difference signal between the input signal and the feedback signal or a signal produced by inverting the feedback signal and adding it to the input signal is inputted to the comparative decider.
The present invention is characterized in that error propagation is detected from the input signal of the above-mentioned comparative decider and from the decision results of the comparative decider, the direction of the error propagation offset is detected from the decision results of the comparative decider, and as a result, the offset of the slice level of the comparative decider is controlled so as to be canceled out, or the DC level of the comparative decider input is controlled so as to cancel out the offset, which suppresses error propagation.
A first preferred aspect of the present invention is characterized in that, if we let the input signal of the above-mentioned comparative decider be y(k) and the output signal of said comparative decider be a binary signal a(k) expressed as ±1, then an error signal ev(k) expressed by ev(k)=[y(k)−Ideal y(k)]·sign[a(k)] is determined using a(k−1)≠a(k+1) as an error computation condition, and the error signal ev(k) thus determined is checked to see whether it exceeds a specific value. If it is detected that the above-mentioned error signal ev(k) exceeds this specific value, the slice level of the above-mentioned comparative decider is controlled to a corresponding offset value.
A second preferred aspect of the present invention is characterized in that, in the first aspect, the above-mentioned error signal ev(k) is accumulated for a plurality of samples that satisfies the error computation condition a(k−1)≠a(k+1), and if the cumulative value exceeds the above-mentioned specific value, the slice level of the above-mentioned comparative decider is controlled to a corresponding offset value.
A third preferred aspect of the present inventi

Affiliated with

Also associated with

Rating

Error propagation detection and control method, and memory... does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Error propagation detection and control method, and memory..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Error propagation detection and control method, and memory... will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3266290