Error detection/correction and fault detection/recovery – Pulse or data error handling – Memory testing
Reexamination Certificate
1998-02-27
2001-09-18
Decady, Albert (Department: 2133)
Error detection/correction and fault detection/recovery
Pulse or data error handling
Memory testing
C714S704000, C714S794000, C714S795000, C714S796000, C360S053000, C360S031000
Reexamination Certificate
active
06292912
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a disk drive such as a magnetic hard disk drive. More particularly, the present invention relates to such a drive having a built-in self-test system for characterizing performance of the drive.
2. Description of the Prior Art
A huge market exists for bard disk drives for mass-market host computer systems such as servers, desktop computers, and laptop computers. To be competitive in this market, a hard disk drive must be relatively inexpensive, and must accordingly embody a design that is adapted for low-cost mass production In addition, it must provide substantial capacity, rapid access to data, and reliable performance. Numerous manufacturers compete in this huge market and collectively conduct substantial research and development, at great annual cost, to design and develop innovative hard disk drives to meet increasingly demanding customer requirements.
An appreciable portion of such research and development has been, and continues to be, directed to developing effective and efficient ways to conduct, as part of the manufacturing process, unit-by-unit testing of drives. One aspect of such testing relates to determining the effect random noise has on drive performance. Data produced from such testing are useful in tuning processes directed to improving drive performance. Another aspect of such testing relates to determining the existence and location of defects such as defects in the media.
Random noise presents difficulties particularly in circumstances in which its magnitude is material in relation to the magnitude of a signal. In other words, a low signal-to-noise ratio (“S/N”) presents problems. In a drive, a low S/N presents problems in attempting to achieve high areal density. Areal storage density relates to the amount of data storage capacity per unit of area on the recording surfaces of the disks. The available areal density may be determined from the product of the track density measured radially and the linear bit density measured along the tracks.
The available linear bit density depends on numerous factors including the performance capability of certain circuitry that is commonly referred to as a “read channel.” One type of read channel is referred to as a peak-detecting channel; another type is referred to as a sampled-data channel. The type referred to as a sampled-data channel is a category including a partial response, maximum likelihood (“PRML”) channel, a EPR4 channel, and a E
2
PR4 channel.
In a hard disk drive having any of these read channels, the read channel receives an analog read signal from a transducer during a read operation. The analog read signal is characterized by a “channel frequency.” As used in this art, “channel frequency” is the reciprocal of a time period “T,” where the “T” is the time period consumed while an elemental-length magnet passes under the transducer during a read operation with the disk spinning at a constant angular velocity. In this regard, the length of each magnet recorded along a track as a result of a write operation is, to a first order of approximation, either an elemental length or an integer multiple of the elemental length. Each elemental length magnet can be referred to as a “bit cell” that is defined during a write operation.
The analog read signal always contains some random noise. The analog read signal, and certain other signals produced by processing the analog read signal and that also contain noise, are referred to herein as noise-corrupted signals. One such other noise-corrupted signal is a signal produced by filtering the analog read signal by means of a low-pass filter. Such filtering may reduce but not eliminate noise, and the filtered signal is also noise corrupted. Further signal processing in the read channel provides for producing a digital signal comprising detected symbols, any of which can be in error in representing recovered data. Such a digital signal is referred to herein as an error-prone signal.
In a hard disk drive employing a peak detecting channel, digital data are represented in the media by transitions between oppositely magnetized bit cells. Provided that the transitions between oppositely magnetized bit cells do not unduly interfere with each other, each such transition causes a peak in the analog read signal, and a peak-detecting channel employs a peak detector that detects such peaks, and produces digital signal in the form of a serial, binary-valued signal that is an error-prone signal for numerous reasons. One reason why the peak detector produces an error-prone signal is random noise; this source of error presents a problem for any type of channel. Another reason relates to interference between adjacent transitions. Interference between such transitions is referred to as intersymbol interference and adversely affects performance of a peak detecting channel increasingly as a function of channel rate.
A sampled-data channel employs sampling circuitry that samples a noise-corrupted analog read signal to produce a sequence of noise-corrupted samples. The samples so produced are provided in sequence to a detector. Such a detector may be organized such that its detection decisions are based on a sequence of the samples. Such a detector is sometimes called a “maximum likelihood sequence detector.” A so-called “Viterbi detector” is an example of a maximum likelihood sequence detector. In a sampled-data channel employing a Viterbi detector, circuitry responds to the noise-corrupted samples to produce error-prone symbols and the produced error-prone symbols are mapped to binary-valued error-prone symbols. In a PRML channel, such internally-produced error-prone symbols are often referred to as: “−1”; “0”; and “+1”; and the binary-valued error-prone symbols are supplied to a deserializer to produce a parallel-by-bit digital signal.
Prior art methods for characterizing the performance of a disk drive are time consuming, costly, and inefficient. Prior art methods include various ways to perform test operations to produce either a bit error rate (“BER”) or a histogram of noise magnitudes.
A BER of 10
−x
means that, on the average, there is no more than one error per 10
−x
bits. A raw BER for a disk drive is typically in the range of 10
−6
BER to 10
−10
BER. The raw BER is estimated without using an ECC correction system to correct errors in a data sequence. A user BER is usually lower than the raw BER and is improved using the ECC correction system.
The BER can be used for fine tuning electronic components in the disk drive. The BER test is commonly repeated after tuning the electronic components.
Prior art methods for estimating the BER require a protracted read operation that involves reading a large number of samples and counting the number of bit errors that occur during the read operation. Prior art methods commonly require reading 10
8
samples to produce a reasonably precise estimate of BER when BER is in the neighborhood of 10
−6
BER This is time consuming and inefficient. Other prior art methods require using large and expensive test equipment to produce an estimate of the BER. This is costly as well as inefficient for use in the manufacturing environment.
U.S. Pat. No. 4,578,721 discloses a “window margin” method for estimating the bit error rate for disk drives employing peak detection read channels. The “window margin” method is not suitable for disk drives employing sampled data detection channels.
U.S. Pat. No. 5,355,261 discloses a method for estimating a BER for disk drives having a partial response maximum likelihood data detection channel. This method requires comparing read back data bits and known data bits to count read back errors.
A publication titled “A WINDOW-MARGIN LIKE PROCEDURE FOR EVALUATING PRML CHANNEL PERFORMANCE”, IEEE Transactions on Magnetics, Vol. 31, No.2, March 1995, discloses a method for estimating the BER that requires counting read back errors during the read operation.
As for a test for measuring the performance of a d
Baumer Howard Anthony
Cloke Robert Leslie
Lee Patrick James
De'cady Albert
Lamarre Guy
Shara Milad G
Western Digital Technologies Inc.
LandOfFree
Disk drive having built-in self-test system for... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Disk drive having built-in self-test system for..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Disk drive having built-in self-test system for... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2517321