Pulse or digital communications – Receivers – Particular pulse demodulator or detector
Reexamination Certificate
1998-07-30
2002-01-08
Deppe, Betsy L. (Department: 2634)
Pulse or digital communications
Receivers
Particular pulse demodulator or detector
C375S263000, C360S065000
Reexamination Certificate
active
06337889
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention is generally related to a signal processing system for either a magnetic disk apparatus or an optical disk apparatus. More specifically, the present invention is directed to a high-efficiency demodulating method of a high-order partial response system such as an EEPRML (Extended Extended Partial Response Maximum Likelihood) signal processing system and an EEEPRML (Extended EEPRML) signal processing system.
In magnetic disk apparatuses, the partial response maximum likelihood (will be abbreviated as a “PRML” hereinafter) signal processing system combining the partial response class
4
(PR
4
) and the maximum-likelihood decoding system are practically available as a high-efficiency signal processing system. A high-efficiency signal processing system implies a system capable of realizing a desirable data error rate at a low SIN (signal-to-noise) ratio. Very recently, such high order partial response systems have been practically utilized as signal processing systems capable of reproducing signals at SIN ratios lower than that of the PRML system, for instance, the EPRML system by combining the EPR
4
(Extended PR
4
) with the maximum (most) likelihood decoding system.
FIG. 1
shows a structural example of the construction of a general magnetic disk apparatus using a PRML signal processing system. Original data is supplied to an error correcting encoder
7
through an interface circuit
8
so that the original data is added with redundant data necessary for error correction. Next, the original data added with redundant data is subjected by a data modulator
6
to modulation necessary for the PRML system and is recorded on a magnetic disk
3
by a magnetic head
4
through a recording/reproducing amplifier
5
. A signal reproduced from the magnetic disk
3
is passed through the recording/reproducing amplifier
5
and then PRML-processed by a data demodulator
1
. The demodulated data is error-corrected by an error correcting decoder
2
and is thereafter converted through the interface circuit
8
into the original data. The operation and the arrangement of this data modulator
6
and data demodulator
1
will now be explained in more in detail with reference to
FIG. 2
indicating a relationship between a magnetic recording/reproducing system and a partial response system. A first description will now be made of process operations executed on the data recording side. The data outputted from the error correcting encoder
7
is penetrated through a precoder
9
constructed of a delay element and modulo (Mod.
2
), and then is recorded via a recording amplifier S on a recording medium. This precoder
9
is employed so as to prevent erroneous propagation of data which is caused during the demodulating operation.
Next, a description will now be made of a processing operation on the reproducing side. The magnetization on the recording medium is reproduced as a waveform having a differential characteristic by the recording/reproducing head. PR
4
may regard this differential characteristic as a differential system of (1−D). In this case, symbol “D” implies a 1-bit delay calculator. The reproduced waveform is supplied to the equalizer
10
so as to be equalized in such a manner that a response of the waveform becomes (1+D). As a result, a total transfer characteristic in the output of the equalizer becomes (1−D
2
). Thereafter, a data discrimination of the data is carried out in the maximum decoder
11
.
FIG. 3
represents a response of a regenerative isolated waveform (note that a step response will be simply referred to as an “isolated waveform” hereinafter) in the case that a step waveform is magnetically recorded. PR
4
implies that the isolated waveform is regarded as a waveform enlarged to 2 time slots, as indicated in FIG.
3
A. This waveform owns such a characteristic having (1+D). Also, as indicated in
FIG. 3B
, EPR
4
implies that the isolated waveform is regarded as a waveform enlarged to 3 time slots. This waveform owns such a characteristic having (1+D)
2
. Furthermore, as indicated in
FIG. 3C
, EEPR
4
implies that the isolated waveform is regarded as a waveform enlarged to 4 time slots. This waveform owns such a characteristic having (1+D)
3
.
While considering the EEPR
4
system as an example, the high order partial response system will now be summarized.
A total transfer characteristic of EEPR
4
constitutes (1−D)•(1+D)
3
as a product of a transfer characteristic of an isolated waveform and another transfer characteristic of a magnetic recording system. An impulse response of the EEPR
4
system determined by this product is represented in FIG.
4
. As apparent from a waveform “a” shown in
FIG. 4
, an isolated waveform of EEPR
4
owns amplitude characteristics (normalized ratio) of 1, 3, 3, 1 every bit period. As a consequence, as indicated in a waveform “b” of
FIG. 4
, a response of an isolated pulse is obtained by superimposing the isolated waveforms inverted along the upper/lower directions with each other by shifting a 1-bit time period. In other words, the response of the isolated pulse becomes 1, 2, 0, −2, −1. In
FIG. 5
, there is shown a trellis diagram of EEPRML obtained by combining a maximum likelihood decoder with EEPR
4
. As is well known in this field, the operation of the EEPRML system may be explained based upon the trellis diagram. In the drawing, symbol “ak” indicates an input signal to EEPRML at a time instant “k”. In this case, reference numeral
12
indicates a state, and reference numeral
13
shows a state transition. An upper stage of a label (ak/yk) and a lower stage thereof indicate an input signal value and an output signal value, respectively. The states of the respective signal processing systems are determined by the past input signal series. In EEPRML, a level of a reproduction signal at the present time instant is influenced by signals over the past 4 time slots. Assuming now that a state at a time instant “k” is equal to “sk”, it is given as
sk=((ak−4, ak−3, ak−2, ak−1)1ak(1,0)), and a total number of states becomes 16. At a time instant “k−1”, state transitions originated from a plurality of states are collected to a specific state at the time instant “k”. With respect to these state transitions, a squared value of a difference between an output signal and an input signal, which are indicated at a low stage of each label, will be referred to as a “branch metric”. Also, an accumulated value of branch metrics until the present time instant with respect to each of the states will be referred to as a “path metric”. Among the state transitions collected to a specific-state at the time instant “k”, only such state transitions that a summation becomes a minimum value and this summation is made from path metrics until a time instant “k−1” and branch metrics corresponding to the respective state transitions are selected as a state transition (path) capable of satisfying a maximum likelihood condition (most certain). This stage may be subdivided into the below-mentioned steps. In other words, the path metrics are added to the branch metrics (Add). Next, these added values are compared with each other every state, and such a state transition which becomes a minimum value is selected (Select). A series of these operations will be abbreviated as an “ACS”. The maximum (most) likelihood decoding method is such well-known method that this ACS operation is repeatedly performed at each time instant and under each state, and then when the path metrics are finally converged onto one path on the trellis diagram, the data is determined.
The performance of EEPRML is determined by a minimum free distance (Dfree). In this case, “Dfree” implies a minimum difference of path metrics among various sorts of combinations from a specific node to another specific node on the trellis diagram shown in FIG.
5
. It is known that “Dfree” of EEPRML is equal to 6. Furthermore, distances between signals subsequent to “Df
Kobayashi Naoya
Kondo Masaharu
Mita Seiichi
Sawaguchi Hideki
LandOfFree
Partial response demodulating method and apparatus using the... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Partial response demodulating method and apparatus using the..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Partial response demodulating method and apparatus using the... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2820020