Honeywell Information Systems Inc.

Donald J Rathbun

Bruce H Tarbox

Edward F Getson, Jr

Albert T McLaughlin

John H Kelley

Taian Su

David B O'Keefe

Douglas A Topliffe

Michael J Funke

Peter P Campbell

A disk drive of a mass storage subsystem includes areas on a disk surface wherein a vendor-generated defective sector log, a software-generated defective sector log and an alternate sector log are stored. A random access memory (RAM) stores a copy of the defective sector logs. During a seek operation, firmware tests the defective sector logs in RAM to generate the alternate sector log for that cylinder number. During the read or write operation, the alternate sector log is checked before processing the sector to determine if it is a defective sector. If the sector is defective, the head is positioned to another cylinder at a head and sector address read from the alternate sector log.

Mass storage disk drive defective media handling

A track on a disk surface of a disk drive is formatted in sectors, each sector having an address portion and a data portion. The disk drive generates a byte clock signal which increments a counter. The counter output signals address a read only memory which generates signals to control the address comparison in the address portion and the reading or writing of data bytes in the data portion of the sector.

Disk subsystem read write control

A track of a disk device is formatted on a single revolution of the disk by using a read only memory (ROM) to store control codes and a random access memory (RAM) to store address field and data field bytes. A DMA controller simultaneously addresses ROM and RAM. Control codes are read into a control first in-first out memory and data codes are read into a data first in-first out memory. The control codes are applied to a decoder whose output signals control cyclic redundancy check and error detection and correction logic as well as the data first in-first out memory. The serial output from both the data first in-first out memory and the cyclic redundancy check logic are written on disk track.

Single revolution disk sector formatter

A diskette read data recovery system generates a clock which is locked to an incoming data stream. In a phase locked loop, a signal generated by an oscillator and frequency dividers is compared in phase to the incoming data stream to provide first or second signals depending on whether the incoming data signal leads or lags the clock signal. In order that the system may handle different types of incoming signals, different frequency divider circuits in the phase locked loop are selected for different incoming signals.

Diskette read data recovery system

Digital data is recorded on the surface of a magnetic media such as a disk or diskette in the form of magnetic flux transitions identifying clock and data information in either a frequency modulation (FM) mode or a modified frequency modulation (MFM) mode. A read head senses the flux transitions which are in turn converted to digital signals. Apparatus converts the time between successive digital signals into PROM addresses. The PROM address locations store coded signals identifying the data as binary ONE or binary ZERO signals.

Digital frequency modulation and modified frequency modulation read

Digital data is recorded on the surface of a magnetic media such as a disk or diskette in the form of magnetic flux transitions identifying clock and data information in a modified frequency modulation (MFM) mode. A read head senses the flux transitions which are in turn converted to digital signals. A counter in the adapter starts to count when the adapter receives a digital signal. The count is transferred to a register and the counter presets when the adapter receives the next digital signal. The count is indicative of the time between the successive digital signals and should be representative of multiples of an integer. The count signals stored in the register address a read only memory whose output signals preset the counter to a value to compensate for the difference between the expected time and the actual time between the successive digital signals thereby reducing the read error rate.

Digital read recovery with variable frequency compensation using read

Information to be written in the form of magnetic flux reversals on the surface of a disk or diskette is applied in serial fashion to a first shift registor. The parallel outputs of the shift register address a PROM. The PROM output is applied to a second shift register in the form of clock and data bits to be written on the disk or diskette magnetic surface. Control signals applied to the PROM address terminals select the mode, FM or MFM, the address mark or if precompensation is required.

Write precompensation and write encoding for FM and MFM recording

A logic system requiring no tuning adjustments is provided for converting an MFM encoded information stream read from a mass storage medium to a non-return-to-zero (NRZ) information stream. The MFM encoded information stream is routed through an input shift register to provide plural information bit cells in parallel. Outputs of the shift register are sampled with a multiplexer to generate timing strobes for detecting an address mark, and for identifying clock bits, data bits and logic zero data appearing in the MFM encoded data field following the address mark. Clock bits are separated from the data, and both data bits and logic zero data are applied serially to an output shift register to form a serial NRZ data stream. Each time a data bit or logic zero data is loaded into the output shift register, a synchronization strobe is generated to transfer the NRZ data to succeeding systems.

Double density read recovery

In a peripheral controller of a data processing system having a plurality of system units electrically coupled to a common communication bus for asynchronous intercommunication, an array of counters responsive to both hardware and firmware are connected in a manner to form a serial control data path. Prior to a data transfer, a serial data stream including an offset range count, a range count and a main memory address is shifted through the counters under firmware control. During a data transfer, the firmware enables the hardware control to increment the memory address and decrement the range count to accommodate the higher data transfer rates characteristic of hardware control.

Range count and main memory address accounting system

A logic system is provided for precompensating data and clock bits of a formatted binary information stream during a modified frequency modulation (MFM) encoding for recording on a magnetic medium. The binary information stream is formatted into a gap field, an address preamble field, an address mark field and a data field. Clock bit generation is inhibited during the gap and address preamble fields. Further, a second of three clock bits occurring during the high order half-byte of the address mark field is suppressed to provide a modified MFM (M.sup.2 FM) field. An address mark is provided thereby for indicating the near proximity of a data field. Beginning with the low order half-byte of the address mark field, both MFM clock precompensation and MFM data precompensation is applied as required. The amount of peak shift occurring in the MFM encoded information stream after precompensation is substantially reduced.

Write precompensation system

In a data processing system wherein a plurality of functional units are interconnected by way of a common communication bus in an environment of high data transfer rates, a logic control system is provided for interjecting firmware control during a data transfer between a disk device and main memory to accommodate unsolicited bus requests without incurring data errors or compromising the data transfer rate. Data transferred between the disk device and a disk controller interfacing directly with the common bus is routed through a FIFO (first-in-first-out) buffer under hardware control. The buffer signals the absence of data in its input register and the presence of data in its output register. The signals are logically combined and ANDed with a firmware controlled logic gate to indicate the occurrence of data transfer states. During such transfer states, data is transferred under hardware control between the FIFO buffer and main memory. When the input register of the FIFO buffer is filled during a data transfer from main memory to the disk device, or when the FIFO buffer is empty during a transfer of data from the disk device to main memory, hardware controlled data transfers are not required. In that event the firmware control system is permitted to access the common bus to service unsolicited bus requests.

Synchronization control system for firmware access of high data rate

A logic data control system including a first-in-first-out (FIFO) buffer predictor is provided for the transfer of data between a main memory unit and a peripheral control unit of a data processing system. Data from main memory is stored into the input registers of the peripheral unit, and thereafter loaded into an array of data FIFOs for transfer to a peripheral storage device. A predictor FIFO operates in parallel with the data FIFOs, and is loaded with a dummy or flag byte each time a data request is made to main memory. When a data word is loaded into the data FIFOs, the input register of the predictor FIFO is sensed. If the flag byte in the predictor FIFO has dropped from the input register into the FIFO stack, a request is issued to main memory for an additional data word. When the data FIFOs are filled, the predictor FIFO also is filled and cannot generate an additional data request until a data byte has been unloaded from the data FIFOs to a peripheral storage device. The input register to the predictor FIFO thereupon is emptied, and another data request may be made to main memory.

FIFO look-ahead system

Data and clock bits are sampled by use of two delay means such as shift registers which are enabled at a clock rate which is set so that a first such delay means coupled to receive the bits includes no more than one bit therein at any time. The receipt of a bit substantially at the midway point of the second delay means, which is coupled serially to receive bits from the first delay means, causes a sampling signal to be generated in response to which a determination is made as to whether another bit has been received by the first delay means. Further logic is provided to recover the data bits.

Growing community of inventors

Andover, MA, United States of America

Donald J Rathbun