A C - 2 1 0 0 0 C A V I A R WESTERN DIGITAL NO MORE PRODUCED Native| Translation ------+-----+-----+----- Form 3.5"/SLIMLINE Cylinders | 2100| | Capacity form/unform 1083/ MB Heads 4| 16| | Seek time / track 11.0/ 3.0 ms Sector/track | 63| | Controller IDE / ATA2 FAST/ENHA Precompensation Cache/Buffer 128 KB ADAP./SEGMEN Landing Zone Data transfer rate MB/S int Bytes/Sector 512 16.600 MB/S ext PIO4 Recording method GCR8/9PRML operating | non-operating -------------+-------------- Supply voltage 5/12 V Temperature *C 5 55 | -40 60 Power: sleep W Humidity % 8 80 | 5 95 standby W Altitude km -0.305 3.048| -0.305 12.192 idle W Shock g 10 | 150 seek 7.0 W Rotation RPM 5200 read/write 5.1 W Acoustic dBA 37 spin-up 12.4 W ECC Bit REED SOLOMON MTBF h 300000 Warranty Month 36 Lift/Lock/Park YES Certificates CE,CSA,FCC,IEC950,TUV,UL1950
WESTERN CAVIAR SERIES INSTALLATION GUIDE 79-000564-007
+---------------------------------------------------------+ | |XX | |XX J2 | |XX Inter- | |XX face | |XX | |.X | |XX | |XX | |XX | |XX | |X1 | |+-+ | || |J8 | |+-1 | |XX Power | |XX J3 +---------------------------------------------------------+ 1
J2 J8 J3 +39------------------------------------1++5-3-1++-------+ |o o o o o o o o o o o o o o o o o o o o||o o o||O O O O| |o o o o o o o o o o o o o o o o o o o||o o o||4 3 2 1| --+40------------------------------------2++6-4-2+++-+-+-++---- | | | +12V (Pin 20 keyed) | | +- GND | +--- GND +----- +5V
WESTERN AC21000 TECHNICAL REFERENCE MANUAL
J8 Master/Slave/Cable Select Configuration
+5-3-1+ Single (Neutral Position) |xxx o| Factory default. The jumper in this position has no effect |o o o| on single hard drive configurations. +6-4-2+
+5-3-1+ Single Drive +5-3-1+ Master Drive |o o o| Configuration |X o o| Configuration |o o o| |X o o| (Dual Drives) +6-4-2+ +6-4-2+
+5-3-1+ Slave Drive |o X o| Configuration |o X o| (Dual Drives) +6-4-2+
The Caviar can be assigned as either a single, master, or slave
Caviar drives are shipped with a jumper shunt in the neutral storage
position (across pins 5 and 3).
DUAL DRIVE OPTION
The Caviar supports ATA-2 dual drive operations by means of
configuration options for master or slave drive designation. The
Caviar is 100% ATA-2 compatible regarding the timing of the PDIAG and
DASP signals. A jumper must be placed in the drive's option area for
both master and slave configurations. If a jumper is placed on the
cable select (CSEL) option, the drive address selection will be
determined by the CSEL signal on the drive cable. Connection to the
host is implemented by means of a daisy-chain cable assembly.
The SDH Register contains the master/slave select bit for the Caviar.
The DASP signal is a time-multiplexed indicator of "Drive Active or
Slave Present" on the Caviar's I/O interface. At reset, this signal
is an output from the slave drive and an input to the master drive,
showing that a slave drive is present. For all times other than
reset, DASP is asserted at the beginning of command processing and
released upon completion of the command. If the master drive option
has been configured, the Caviar will not respond to commands or
drive status on the interface when the slave bit is selected in the
The Caviar drive has a jumper block (J8) located next to the 40-pin
connector on the drive. The Caviar can be assigned as either a
single, master, or slave drive.
Caviar drives are shipped with a jumper shunt in the neutral storage
position (across pins 5 and 3).
Single Drive Mode - If you are installing the Caviar drive as the
only hard drive in the system, leave the jumper in the neutral
storage position. Jumpers are not required for single drive
installations. Note that even with no jumper installed, the Caviar
checks the DRIVE ACTIVE/SLAVE PRESENT (DASP) signal to determine if a
slave IDE drive is present. If you have a dual installation (two
hard drives), you must designate one of the drives as the master and
the other as the slave drive. The jumper pins on the J8 connector
need to be configured for the dual installation.
Master Drive Mode - To designate the drive as the master, place a
jumper shunt on pins 5-6. With the Caviar configured as the master
drive, the Caviar assumes that a slave drive is present. The jumper
on pins 5-6 is optional if the slave drive follows the same protocol
(Common Access Method AT Bus Attachment) as the Caviar.
Slave Drive Mode - To designate the drive as the slave, place a
jumper shunt on pins 3-4. When the Caviar is configured as the slave
drive, the Caviar delays spin up for three seconds after power-up
reset. This feature prevents overloading of the power supply during
J3 DC Power and pin connector assignments ------------------------------------------- +------------+ pin 1 +12 V | 4 3 2 1 | pin 2 GND +------------+ pin 3 GND pin 4 + 5 V
WESTERN AC21000/21200/31600 TECHNICAL REFERENCE MANUAL 79-860021-000
Notes On Installation
horizontally vertically +-----------------+ +--+ +--+ | | | +-----+ +-----+ | | | | | | | | | +-+-----------------+-+ | | | | | | +---------------------+ | | | | | | | | | | | | | | | | | | +---------------------+ | +-----+ +-----+ | +-+-----------------+-+ +--+ +--+ | | | | +-----------------+
The Caviar can be mounted in the X, Y, or Z axis depending upon the
physical design of your system. It is recommended that the drive be
mounted with all four screws grounded to the chassis.
Screw Size Limitations
The Caviar is mounted to the chassis using four 6-32 screws.
Recommended screw torque is 5 in-lb. Maximum screw torque is 10
Caution: Screws that are too long will damage circuit board
components. The screw must engage no more than six threads (3/16
inch). Side mounted screws should engage a maximum of .188 inches
(3/16"). Bottom mounted screws should engage a maximum of .250
It is recommended that the drive be mounted with all four screws in
the side grounded to the chassis. The drive must be grounded with at
least one mounting screw. Side mounting: Use four metal screws.
Top face mounting: Use four metal screws.
Determining Your Configuration ------------------------------ You can configure the Caviar in one of two ways: 1. The drive is cabled directly to a 40-pin connector on the motherboard, or 2. The drive is cabled to an adapter card mounted in one of the expansion slots in the computer.
Both configurations use a 40-pin host interface cable.
If you are using the Caviar drive as one of two hard disk drives in
the computer (dual installation), you may use either configuration.
In dual installations, you must use a 40-pin host interface cable
with three connectors and daisy-chain the two drives to the
motherboard or adapter card.
Dual installations require a master/slave drive configuration, where
one drive is designated as the primary (master) drive and the other
is designated as the secondary (slave) drive. The Caviar drive is
compatible in dual installations with other IDE drives that support a
Mounting the Drive
For dual installations, it is usually easier to completely install
one IDE drive in the lower position first. The order of IDE drives is
unimportant if you are using two Western Digital drives. As explained
previously, one must be jumpered as the master drive and the other as
the slave drive. When installation is complete, the drives are
Cabling and Installation Steps
Make sure your interface cable is no longer than 18 inches (including
daisy chaining) to minimize noise that is induced on the data and
control buses. When connecting two drives, use a daisy-chain cable
that has three 40-pin connectors. Connectors should be placed no more
than six inches from the end of the cable. If only one drive is
connected, it should be placed on the end of the cable.
Caution: You may damage the Caviar drive if the interface cable is
not connected properly. To prevent incorrect connection, use a cable
that has keyed connectors at both the drive and host ends.
Pin 20 has been removed from the J2 connector. The female connector
on the interface cable should have a plug in position 20 to prevent
incorrect connection. Make sure that pin 1 on the cable is connected
to pin 1 on the connectors.
The order in which you perform the following steps will vary
depending on your system.
1. Attach the end of the 40-pin interface cable to the 40-pin J2 connector on the back of the Caviar hard drive. For dual installations, connect the two drives together by using a three-connector interface cable. Match the orientation of pin socket 1 on the 40-pin IDE cable to pin 1 on the connector.
2. Thread the cable through the empty drive bay and slide in the Caviar drive.
3. Mount the Caviar drive in the drive bay using four 6-32 screws. Be sure to use the correct size screws. Do not install the screws past six threads (3/16 inch). Screws that are too long will damage the Caviar drive.
For proper grounding be sure to use ALL four screws.
WESTERN AC21000/31600 TECHNICAL REFERENCE MANUAL 1/96
The Caviar AC21000 and AC31600 Enhanced IDE (EIDE) disk drives are
high-performance solutions designed to meet the requirements of
today's most powerful systems from home and business PC's to
workstations and servers. These drives are based on our successful
proven design concepts. By combining enhanced electronics with new
head and read-channel technology, we have produced the market-leading
areal density and the highest performance Caviar drives to date.
High-speed host data transfers, advanced caching, increased
rotational speeds, and low mechanical latency combine to give
the AC21000 and AC31600 the level of performance demanded by today's
most advanced systems.
These drives support host data transfers of 16.6 MB/s Mode 4 PIO and
16.6 MB/s Mode 2 multi-word DMA enabling VESA VL or PCI local bus
EIDE integration. The AC21000 and AC31600 offer a rotational speed of
5200 RPM and include CacheFlow4.
CacheFlow4 offers adaptive read and write caching capabilities which
complements the advanced caching capabilities of today's major
operating systems. An average read seek time of sub 10/11 ms and
rotational latency of 5.76 ms combine to provide fast mechanical
access. The AC21000 and AC31600 offer performance beyond that of the
ISA bus. Optimum performance is obtained when these drives are
integrated into a VESA VL or PCI local bus EIDE environment. System
integration of these drives in a DOS or Windows environment requires
either BIOS, device driver or operating system support for EIDE disk
drives with capacities greater than 528 MB.
The AC21000 and AC31600 drives support advanced power management
capabilities that can reduce power requirements over 85 percent. All
Caviar drives are preformatted (low-level) and defects are mapped
out before shipment. Additional Caviar features include logical
block addressing, linear logical/physical universal address
translation, automatic head parking, embedded servo control data on
each track, and ECC on-the-fly correction.
Western Digital's award-winning Caviar drives are designed and
manufactured to the highest standards of quality and reliability.
This is demonstrated by their three-year warranty, 300,000 hours
Mean Time Between Failure, and guaranteed compatibility.
The Caviar AC21000 and AC31600 drives are today's solution to the
computer market's ever-increasing demand for higher performance and
expanded connectivity capability and they still provide the
advantages of low cost, compatibility and ease of use.
Advanced Product Features ------------------------- - CacheFlow4 - Western Digital's unique, fourth-generation caching evaluates the way data is read from and written to the drive and adapts on-the-fly to the optimum read and write caching methods. CacheFlow4 minimizes disk seeking operations and the overhead due to rotational latency delays.
CacheFlow4 includes random and sequential write cache. Incorporating write cache with other CacheFlow4 features enables the user to cache both read data as well as write data. Multiple writes can now be held in the cache and then written collectively to the hard disk later. Data is held in the cache no longer than the time required to write all cached commands to the disk.
CacheFlow4 constantly evaluates not only the size of the read data request but the type of data request, that is, whether the application is sequential, random, or repetitive.
CacheFlow4 then dynamically partitions the Caviar's 128-Kbyte DRAM buffer into equalsized segments and selects the appropriate caching mode for optimum system performance.
- Advanced Host Transfer - The AC21000 and AC31600 support Mode 4 PIO (16.6 MB/s) and Mode 2 multi-word DMA (16.6 MB/s) as defined by the ATA-2 standards. To achieve Mode 4 PIO burst transfers, hard disk drives must be able to throttle the host via the IORDY signal. Systems typically require a high-speed VL or PCI local bus in order to support Mode 4 PIO.
- High-Speed DMA Capability - DMA Read and DMA Write commands are ATA-2-compatible and provide significant improvement in CPU bandwidth over conventional PIO data transfers. The system CPU is free to accomplish other tasks while the Caviar drive transfers data directly to/from system memory.
- Power Conservation - The AC21000 and AC31600 support the ATA-2 power management command set. This command set allows the host to reduce the power consumption of the drive by issuing a variety of power management commands.
- Extended Contact Start/Stop Capabilities (CSS) - Today's power-managed desktop systems result in more contact start/stop cycles to the disk drive. The AC21000 and AC31600 support a minimum of 60,000 start/stop cycles.
- Zoned Recording - The AC21000 and AC31600 employ Zoned Recording to increase the data density on the outer tracks of the drive. The outermost tracks contain more sectors than the innermost tracks, thereby increasing the total capacity of the drive.
- Block Mode - ATA-2 compatible Read Multiple and Write Multiple commands are supported. Block mode increases overall data transfer rates by transferring more data between system interrupts.
- Logical Block Addressing (LBA) - The AC21000 and AC31600 support both LBA and CHS-based addressing. LBA is included in advanced BIOS and operating system device drivers and ensures high-capacity disk integration.
- Automatic Head Parking - Head parking is automatic with Caviar drives. On power down, the heads retract to a safe, non-data landing zone and lock into position, improving data integrity and resistance to nonoperational shock.
- Advanced Defect Management - These Caviar drives are preformatted (low-level) at the factory and come with a full complement of automatic defect management functions. Extensively tested during the manufacturing process, media defects found during intelligent burn in are mapped out with Western Digital's high performance defect management technique. No modifications are required before installation.
- Embedded Servo Control - These Caviar drives feature an embedded servo concept as the means of providing sampled position feedback information to the head position servo system. Servo bursts are located along a radial path from the disk center, ensuring that head positioning data occurs at constant intervals. This high sampling rate supports the high frequency servo bandwidth required for fast access times as well as highly accurate head positioning. The embedded servo concept provides the means of generating accurate feedback information without requiring a full data surface as would a dedicated servo control concept.
- Dual Drive Operation - These Caviar drives support dual drive operation by means of a "daisy chain" cable assembly and configuration options for master or slave drive designation.
- Universal Address Translation - These Caviar drives provide a linear disk address translator to convert logical sector addresses to physical sector addresses which provides for easy installation and compatibility with numerous drive types.
- Guaranteed Compatibility - Western Digital performs extensive testing in its Functional Integrity Test Lab (FIT Lab(TM)) to ensure compatibility with all 100% AT-compatible computers and standard operating systems.
- Reed Solomon ECC On-the-Fly - The Caviar implements Reed Solomon error correction techniques to obtain extremely low read error rates. This error correction algorithm corrects errors on-the-fly without any performance penalties. It allows for hardware corrections of up to a 24-bit error span on-the-fly.
- Automatic Defect Retirement - If the Caviar drive detects a defective sector while writing, it automatically relocates the sector without enduser intervention.
Every Caviar undergoes factory-level intelligent burn in, which
thoroughly tests for and maps out defective sectors on the media
before the drive leaves the manufacturing facility. Following the
factory tests, a primary defect list is created. The list contains
the cylinder, head, and sector numbers for all defects.
Defects managed at the factory are sector slipped. Grown defects that
can occur in the field are mapped out by relocation to spare sectors
on the inner cylinders of the drive.
The Caviar is shipped from the factory preformatted (low-level) with
all the known defects mapped out. In order to be compatible with
existing industry standard defect management utility programs, the
Caviar supports the logical format command. When the host issues the
Format Track command, the Caviar performs a logical version of this
command in response to the host's interleave table request for good
and bad sector marking or assign/unassign the sector to/from an
If the host issues the Format Track Command during normal operating
modes, the data fields of the specified track are filled with a data
pattern of all zeros. The Format Track Command can be used to
mark/unmark bad sectors, and reassign unrelocated sectors.
Automatic Defect Retirement
The automatic defect retirement feature automatically maps out
defective sectors while writing. If a defective sector appears,
Caviar finds a spare sector at the inner diameter cylinders and
relocates the sector.
Error Recovery Process
The Caviar has four means of error recovery:
- ECC On-the-Fly
- Read/Write Retry Procedure
- Extended Read Retry Procedure
- Extended (Firmware Assisted) ECC
ECC On-the-Fly - If an ECC error occurs, the Caviar attempts to
correct it on-the-fly without retries. Data can be corrected in this
manner without performance penalty.
Read/Write Retry Procedure - This retry procedure is used by all disk
controller error types. If this procedure succeeds in reading or
writing the sector being tried, then recovery is complete and the
controller continues with the command. Each retry operation also
checks for servo errors. This procedure ends when error recovery is
achieved or when all possible retries have been attempted.
Extended Read Retry Procedure - This retry procedure tries
combinations of positive/negative track offsets, and data DAC
manipulations to recover the data. This retry procedure is applicable
only to read data recovery. The Read/Write Retry procedure is used
to perform the actual retry operation. When an extended retry
operation has been successful, the controller continues with the
command. The controller ensures that any changes in track offset or
data DAC settings that exist are cleared before the command
Extended (Firmware Assisted) ECC - If an ECC error is too large to
correct using ECC on-the-fly, the Caviar can attempt to correct the
error using Extended Error Correction. This allows correction of
large ECC errors that ECC on-the-fly cannot correct. However, the
Extended Error Correction process takes more time than ECC
on-the-fly to return the corrected data.
REED SOLOMON ECC On-the-Fly
The Caviar implements Reed Solomon error correction techniques in
hardware to reduce the uncorrectable read error rate. This allows a
high degree of data integrity with no impact on the drive's
performance. Because on-the-fly corrected errors do not require the
drive's firmware to assist with error correction, they are invisible
to the host system. To obtain the ECC check byte values, each byte
within the sector is interleaved into one of three groups, where the
first byte is in interleave 1, the second byte is in interleave 2,
the third byte is in interleave 3, the fourth byte is in interleave
1, and so on. See Figures 3-2 and 3-3 for examples of interleaving.
Interleaving and the ECC formulas enable the drive to detect where
the error occurs. A maximum of one byte can be corrected in each
interleave without firmware assistance.
Firmware Assisted ECC
With firmware assisted ECC, a maximum of 3 bytes can be corrected in
each interleave. In this case, a 65-bit error span is the maximum
that is always correctable with firmware assistance because the
entire error span will never occupy more than three bytes in each
Universal Address Translation
The Caviar implements linear address translation. The translation
mode and translated drive configuration are selected by using the Set
Drive Parameters command to issue head and sector/track counts to the
translator. Caviar supports universal translation. Therefore, any
valid combination of cylinder, head, and SPT can be assigned to the
drive as long as the total number of sectors is not greater than the
physical limits. The product of the cylinder, head and sectors/track
counts must be equal to or less than the maximum number of sectors
available to the user. The maximum number of sectors per drive for
the AC21000 are 2,116,800. The maximum number of sectors per drive
for the AC31600 are 3,173,184.
Each sector consists of 512 bytes.
The minimum value for any translation parameter is one. The maximum
value for any translation parameter is as follows:
Sectors/Track - 255 Heads - 16 Cylinders/Drive - 65,535
The values in the Sector Count Register and the SDH Register
determine the Sectors Per Track (SPT) and heads. Regardless of the
values of the SPT and the heads, Caviar is always in the translation
The AC21000 and AC31600 support the ATA-2 power management commands
that lower the average power consumption of the disk drives. For
example, to take advantage of the lower power consumption modes of
the drive, an energy efficient host system could implement a power
management scheme that issues a Standby Immediate command when a host
resident disk inactivity timer has expired. The Standby Immediate
command would cause the drive to spin down and enter a low-power
mode. Subsequent disk access commands would cause the drive to spin
up and execute the new command. To avoid excessive wear on the drive
due to the starting and stopping of the HDA, the host's disk
inactivity timer should be set to no shorter than ten minutes.
High-Speed DMA Capability
By engaging an ATA-2 compatible, Mode 2 multi-word DMA, the host CPU
bandwidth is increased because the peripheral data transfer burden is
off-loaded to the system's DMA channel. With the exception of DMA
data transfers, which are limited to Read DMA and Write DMA
commands, all other commands must be performed using PIO. DMA or PIO
data transfer mode selection by the host is performed on a
Advanced Host Transfers
The AC21000 and AC31600 support high-speed Mode 3 and 4 PIO. These
are data transfer modes that utilize hardware handshaking between the
host and the drive via the IORDY signal. When the drive deasserts the
IORDY signal, the host extends the read/write cycle until IORDY is
asserted, thereby eliminating data corruption from overrun and
underrun conditions. When in Mode 3 PIO, data can be transferred in
bursts to and from the host at a rate of up to 11.1 MB per second; in
Mode 4 PIO, the data can be transferred at a rate of up to 16.6 MB
per second. Mode 3 and Mode 4 PIO are enabled on the drive by issuing
a Set Features command. If Mode 3 or Mode 4 PIO is enabled, it can
only be disabled by issuing another Set Features command, a hard
reset, or by cycling power. To support Mode 4 PIO, Flow Control must
be enabled in the host system. If this mode is enabled on a system
that does not support Flow Control, host FIFO errors can occur. Mode
3 and Mode 4 PIO timings were defined to facilitate EIDE drive
integration into VL and PCI local bus systems.
Zoned Recording is a mechanism for increasing the capacity of the
drive by increasing the Bit-Per-Inch (BPI) density of data written
on the longer outer tracks of the drive. Track capacity (number of
sectors) is constant within groups of tracks or zones, and is
increased when the tracks are sufficiently long to accommodate a
significant number of additional sectors. This incremental increase
in track capacity moving outward on the disk surface creates a series
of concentric zones with different data densities.
WESTERN DIGITAL Defect Management Utility
All Caviar IDE drives are defect-free and low level formatted at the
factory. After prolonged use, any drive, including Caviar, may
develop defects. If you continue receiving data errors in any given
file at the DOS level, you can use the defect management utility
WDAT_IDE to recover, relocate and rewrite the user data to the
nearest spare sector and maintain a secondary defect list.
Caution: As with all format utilities, some options in the WDAT_IDE
utility will overwrite user data.
Self-Monitoring, Analysis, and Reporting Technology (S.M.A.R.T.)
S.M.A.R.T. enables a drive's internal status to be monitored through
diagnostic commands at the host level.
The Caviar AC21600, AC32100 and AC32500 drives monitor read error
rate, start/stop count, spin-up retry count, and drive calibration
retry count. All of these attributes are updated and stored on the
hard drive in the reserved area of the disk. The hard drive also
stores a set of attribute thresholds that correspond to the
calculated attribute values. Each attribute threshold indicates the
point at which its corresponding attribute value achieves a negative
WESTERN ENHANCED EIDE
Enhanced IDE Backgrounder
The Computer Market and the IDE Interface:
The computer marketplace is segmented into various classes of
machines divided by user expectations in terms of cost, performance,
compatibility and ease-of-use. The largest distinct segment today is
the personal computer market, characterized by single- user products
supporting a broad user base. The usage of these machines in business
and home environments has dictated an emphasis on cost and
compatibility. Historically, cost and compatibility in the personal
computer marketplace have been more important to mainstream users
than very high performance. The PC user has simply not been willing
to bear the added cost or potential lack of compatibility that
highest performance solutions imply.
Given this criteria, the mainstream volume personal computer market
has standardized on the IDE interface for its primary storage needs.
The success of the IDE interface in the PC market has resulted
primarily from a perfect match between IDE's offerings and the
requirements of the market it serves. Specifically, its low cost of
connection, compatibility, and ease-of-use, compared to alternative
interfaces such as the Small Computer System Interface (SCSI), have
been essential attributes in satisfying an expansive price-sensitive
user group. In addition, because of the broad user base it serves,
the personal computer market has traditionally required only hard
disk support to meet its mass storage requirements. IDE has therefore
evolved as a drive-only interface.
Increasing Need for Performance and Connectivity Flexibility:
As the personal computer market matures, it continues to display an
increased emphasis on enhanced performance and connectivity
capabilities, while maintaining its focus on cost, compatibility and
ease-of-use. The market criteria has therefore grown to include
higher performance attributes without sacrificing the needs of its
price sensitive customers. It is in the realm of higher performance
characteristics and connectivity that today's traditional IDE
interface faces challenges. Other existing interfaces, such as SCSI,
provide greater flexibility and performance options to meet these
requirements, while failing to provide IDE's benefits of
compatibility, cost and ease-of-use.
Western Digital's Enhanced IDE technology addresses the performance
and connectivity challenges facing the IDE interface. Enhanced IDE is
designed to extend the attributes of the IDE interface so that its
characteristics more effectively match the new requirements of the
evolving personal computer market, without forfeiting its traditional
Western Digital and the IDE Interface - Building upon Expertise:
Western Digital's Enhanced IDE technology evolves from the company's
storage expertise within the personal computer marketplace. In 1984,
Western Digital developed the WD1002 floppy and ST506 interface hard
disk controller that IBM utilized in their PC/AT systems. The success
of the PC/AT architecture led to the massive growth of the IBM PC/AT
compatible market. This dramatic growth was in part fueled by WD1002
compatible hard disk controllers and later by Western Digital's
standard-setting WD1003 series of AT controllers.
As the market expanded and became more price sensitive, Western
Digital defined the need for integration of the AT controller
electronics within the disk drive. By working with Compaq Computer
Corporation, Western Digital again drove the technology by proposing
the IDE (Integrated Drive Electronics) interface which was
implemented in the industry's first IDE drive in 1986. The disk
drives used in personal computers have standardized around IDE since
Now, Western Digital continues to lead the industry with its IDE
interface expertise via Enhanced IDE, an approach that expands upon
the existing attributes of the IDE interface and extends its usage
into more demanding environments. Enhanced IDE not only incorporates
high speed host transfer capabilities, support of high capacity disk
drives, and multiple device connectivity, but it also includes
non-disk peripheral support via the Western Digital authored ATAPI
(AT Attachment Packet Interface) specification. This
enhanced IDE-ATA specification enables connectivity of non-disk
peripherals such as CD-ROM and tape drives. The Western Digital
defined ATAPI specification, with participation and endorsement by
key market-making OEMS, CD-ROM suppliers and operating system
suppliers, is yet another example of Western Digital's commitment to
the evolution of the IDE interface.
Enhanced IDE removes many of the existing limitations and issues
associated with the current IDE interface. Removal of these
limitations enables IDE to grow with the industry's increased mass
storage requirements without sacrificing its key cost, compatibility
and ease-of- use attributes. The historical limitations of IDE
relative to other interfaces, such as SCSI, have not threatened IDE's
dominance of the PC marketplace to date. Upcoming personal computer
systems, architected around high performance processors, more complex
operating systems, and more demanding software applications, have
developed storage requirements beyond the realm of today's IDE
capabilities, challenging IDE's dominant role in the PC market.
Specifically, the IDE interface is less flexible and limited in key
areas of performance and connectivity relative to the SCSI interface:
The IDE interface supports two disk drives. The SCSI interface supports multiple devices includingprinters, CD-ROM, tape drives as well as hard disk drives.
The IDE interface is limited to 528MB hard disk capacity as a result of the Int 13h BIOS interface used to access IDE drives. The SCSI interface is not limited in capacity. The IDE interface typically offers 2-3MB/sec host transfer rates on standard ISA bus architected machines. The SCSI interface offers 10MB/sec FAST transfers and up to 20MB/sec FAST/WIDE host throughput.
Western Digital's Enhanced IDE technology offers solutions to the
existing constraints associated with the current IDE interface such
as capacity limitations, slower host transfers, and connectivity
issues associated with the IDE interface and thereby enables a cost
effective, compatible, and easy-to-use interface solution for the
next generation of personal computers.
Components of Enhanced IDE:
Enhanced IDE focuses on removing four primary limitations of the
existing IDE interface. These include:
Removal of the 528MB capacity barrier Breaking the IDE transfer bottleneck Supporting multiple IDE devices Enabling non-disk peripheral connectivity, such as CD-ROM
Below, each of these limitations is discussed and resolved in detail.
Removal of Capacity Limitations
A barrier in implementing IDE disk drives greater than 528MB exists
in today's standard AT system BIOS. This barrier is based on
historical reasons dating from the development of the original AT
machine in 1984. Specifically, it is a limitation of the combined
Interrupt 13 software interface and the IDE interface. The goal is to
change the system BIOS such that this barrier no longer exists,
thereby enabling the usage of high capacity IDE disk drives. Western
Digitial's specification for removing the 528MB barrier is a simple
yet effective method for implementation by BIOS suppliers and system
manufactures who write their own BIOS.
The capacity limitation exists due to the number of bits allocated
for specifying the cylinder, head, and sector address information at
both the Int 13h interface level and at the IDE interface level.
Because Int 13h and IDE specify differing values, combining these two
interfaces produces an artificial 528MB barrier as shown below:
BIOS IDE --------------------------------------------------------------- Limitation Max Sectors/Track 63 225 63 Number of Heads 255 16 16 Number of Cylinders 1024 65536 1024 Maximum Capacity 8.4GB 136.9GB 528MB
Two solutions exist that resolve the existing 528MB barrier problem.
The first method is to have the BIOS translate the CHS address at the
13h interface to the CHS parameters being used at the drive
interface. The Enhanced IDE proposal to break the 528MB barrier is to
utilize the second method of modifying the Int 13h BIOS so that it
translates the cylinder, head, sector information passed to it via
Int 13 into a 28 bit Logical Block Address (LBA). The LBA solution is
believed to be the best method of breaking the 528MB barrier because
it provides a clean and efficient way for future operating system
drivers to access IDE drives.
The LBA translation is loaded into the drive's task file registers.
Bit 6 of the drive's SDH register is set to indicate to the drive's
firmware that it should interpret the information in its task file
registers as LBA rather than cylinder, head and sector information.
This scheme will allow for the full use of all of the bits allocated
for CHS information at the Int 13h interface, thereby supporting up
Using a logical block addressing scheme is attractive primarily
because it is 100 percent compatible with BIOS Int 13 and allows for
reduced overhead, producing higher performance. The logical block
addressing scheme provides the compatibility essential for personal
computer usage as well as enables the implementation of higher
capacity disk drives required for high performance machines.
Western Digital's LBA scheme has been successfully demonstrated by
key system manufacturers writing their own BIOS and by those working
in conjunction with their BIOS suppliers. Systems shipping in
calendar Q4, 1993 will implement this scheme with the Western Digital
Bypassing the AT-IDE Host Transfer Bottleneck:
The ISA bus capabilities are designed to sustain host throughput data
rates of roughly 2-3MB/sec. Relative to SCSI host transfer rates of
5MB, 10MB, and 20MB/sec, the ISA bus is painfully slow for higher
performance applications. Because AT personal computers did not
necessarily demand the higher performance obtained by their
workstation or file server counterparts, 2-3MB/sec wasn't considered
a limiting factor. In addition, the ISA bus capabilities of 2-3MB/sec
did not present a throughput problem because data rates coming off
the media were roughly only 5Mbits/sec, and not a challenge to the
As disk drive areal density technologies progressed, media data rates
began to exceed the 2-3MB/sec ISA host throughput. Buffering either
on the system or the drive was necessary to maintain performance. The
industry's most recent drive offerings far exceed the ISA bus host
throughput by providing media data rates of up to 48Mbit/sec. Due to
these factors, increased buffering is not a cost effective
alternative to faster host throughput.
Fast PIO Transfers:
Other peripherals within the computer, such as video, resolved their
throughput problems via local bus architectures providing a potential
path for improved performance. IDE local bus solutions, leveraged
from the success of video local bus, began appearing in 1992, as a
way to enhance data throughput. These solutions mapped the IDE data
port to the local bus, bypassing the ISA bus and enabling the
maximization of throughput from the media to the drive buffer, on to
the host. These solutions were still not competitive with Fast SCSI
(10MB/sec) due to the "blind" transfer nature of the PIO transfers.
"Blind PIO" transfers indicate host control of data throughput with
the host requesting data (master) and the drive responding (slave).
With blind PIO transfers, the host is unaware or "blind" when
buffered drive bandwidth is 100% available for host transfers.
Because there are cases when only a percentage of bandwidth is
available, blind PIO host requests for data from the drive are based
on the worst case bandwidth availability. This means that even when
the ISA bottleneck is bypassed by connection directly to the local
bus, inability to utilize 100% drive bandwidth prevents full
optimization of host throughput.
Enhanced IDE incorporates an operation called "Flow Control Using
IORDY" (I/O Channel Ready) which allows the drive to "throttle" the
host when necessary and enable burst transfers to take advantage of
100% of the bandwidth. Flow Control thereby gives control of the data
transfer to the drive and eliminates the inefficiencies of blind PIO
by setting the host to maximum drive bandwidth support. This means
that when 100% drive bandwidth is available, the drive will take
control and transfer data to the host.
This operation, based on approved Mode 3 PIO timings of 180ns cycle
times from the Small Form Factor Committee, supports transfer rates
up to 11MB/sec competitive with FAST SCSI solutions. Flow Control is
enabled on the drive by the host issuing a Set Features command, so
that both the host and drive side support this operation. Western
Digital's 540MB drives (shipping beginning September, 1993) support
flow control using IORDY and will be implemented into machines that
take advantage of this feature via low cost ASICS whose functionality
will later be incorporated into core logic chipset solutions.
Although PIO is the standard transfer method supported by the
industry and presents no incompatibility issues (see footnote),
another transfer option exists that provides incremental transfer
benefits beyond PIO. Direct Memory Access (DMA) is based on data
transfer directly to memory rather than via the CPU. DMA transfers
are "throttled" and therefore have historically offered the benefit
of maximizing data throughput. The throttling mechanism associated
with DMA has historically enabled improved data transfers relative to
Type B DMA was defined with the arrival of Extended Industry Standard
Architecture (EISA), and is specified at 4.0MB/sec transfer rates
offering an advantage to the standard 2-3MB/sec PIO data rates.
Although this is an improvement to the standard ISA bus timings, Type
B DMA remains uncompetitive with FAST SCSI timings of 10MB/sec.
With the advent of local bus solutions, a new DMA transfer has
emerged in conjuction with PCI. Type F DMA is defined to support
8.33MB/sec and 6.67MB/sec data rates, a large improvement over Type B
DMA. In conjunction with chipsets capable of supporting 6.67MB and
8.33MB/sec data rates, the Small Form Factor Committee has approved a
new multiword DMA Mode 1 timing specification of a 150ns cycle time.
This enables DMA transfers up to 13MB/sec for future data rate
improvement by allowing multiple words to be transferred for any
given request command. PCI chip sets will be shipping with both EISA
(Type B) and ISA (Type F) configurations in the calendar CYQ4'1993
PIO versus DMA:
The disadvantage of DMA transfer operations is that the PC/AT hard
disk controller and later IDE, evolved around PIO data transfers.
Therefore, the system Int 13h BIOS and the embedded operating system
device drivers have supported PIO transfers versus DMA transfers.
This simply means that BIOS changes and external device drivers are
necessary to achieve the incremental performance that DMA offers.
Western Digital's Enhanced IDE program supports system manufacturers'
choice of either PIO transfers via Flow control with IORDY for Mode 3
PIO data rates or DMA transfers (both Type B and Type F) via the
development of external DMA device drivers supporting Western Digital
hard disk drives. Product platforms based on both high speed transfer
options will be in production in calendar fourth quarter 1993.
Supporting Multiple IDE Devices:
The original IBM PC/AT defined support for two hard disk controllers
and allowed support for up to four disk drives via a primary and
secondary controller. The original BIOS and operating system drivers,
however, only supported the primary controller, limiting the standard
PC configuration to two disk drives. Today's operating systems now
offer both primary and secondary controller support providing an
opportunity to extend peripheral attachment capabilities with IDE.
The addition of a second connector via a hardware change is a simple,
low cost solution that allows for multiple IDE peripheral
The cost of a second IDE connector is less than $1.00. Most core
logic and Super I/O devices have already integrated the capability to
support either the primary or secondary address decode logic and
therefore the cost of the secondary port is simply the 40 pin
connector and surrounding transceivers and resistors. For $1.00, dual
IDE connectors offer support for four IDE devices and satisfy the
expansion needs of the majority of the mainstream personal computer
market, a very cost effective alternative to connectivity via SCSI.
Western Digital's Enhanced IDE program works with system
manufacturers to understand the BIOS implications of a secondary
channel for support of two additional IDE devices. The BIOS must be
able to determine the physical location of the drive based on the Int
13h drive number . Since DOS 3.0 and later will support up to seven
disk drives, only the system BIOS Interrupt 13h needs to be modified
to support primary and secondary IDE. Windows 3.1 accesses the disk
via Interrupt 13h calls to the BIOS. Again, all that is required is
modification to the system BIOS to support dual channel IDE. IBM OS/2
2.0 and 2.1 as well as MS/IBM OS/2 1.31 all support four IDE drives
on dual IDE connectors via their drivers. Netware is hardcoded to
support four IDE connectors or 8 IDE devices. Dual channel IDE
support will be in the final release of Windows NT.
Dual channel IDE not only enables the cost effective and easy
implementation to support multiple disk drives, it presents the
opportunity to expand IDE into non-disk peripheral support. A slow
speed channel and a high speed channel can be developed for efficient
implementation of storage solutions via high performance hard disk
drives and mass data storage vehicles such as CD-ROM and tape drives.
Enabling Non-disk Peripheral Connectivity:
The upcoming high performance desktop machines are demanding
additional storage peripheral support beyond hard disk drives.
Specifically, CD-ROM and tape drives will demonstrate rapid unit
growth rates as these peripherals become a more standard part of the
desktop's configuration. Today's CD- ROMs and tape drives have
multiple interfaces that present compatibility and performance
issues. Development of a standard IDE interface for both CD-ROMs and
tape drives solves cost, compatibility, performance, and ease-of-use
issues in conjunction with enabling the attachment of non-disk
devices via the IDE interface.
Western Digital, with its AT interface expertise, has taken the
leadership position in expanding the IDE interface to support
non-disk peripherals by authoring the AT Attachment Packet Interface
(ATAPI). The specification defines a standard method for interfacing
to a CD-ROM drive (and other non-disk devices) utilizing the existing
ATA host computer hardware and cabling. ATAPI supplements the
definitions of an ATA mass storage peripheral found in the ATA
specification and is compatible with existing ATA hardware without
any changes or additional pins.
Traditional computer architecture has used a register based transport
mechanism. Modern architectures now use packet-based transport
mechanisms. ATAPI is an enhancement to IDE that follows this trend.
Benefits of including a packet-based scheme means adding very few IDE
operation codes. The ATAPI specification adds only a single new IDE
command to obtain functionality and only two additional new IDE
commands to address compatibility. Once a packet-based interface was
defined, the next issue was deciding what command packets definitions
to utilize. Given widespread support for SCSI within peripherals and
within existing operating systems, it was decided to derive ATAPI
command packets from SCSI to minimize development time and expense.
The ATAPI specification is being reviewed by an industry working
group that consists of market-making system manufacturers, CD-ROM
suppliers, silicon designers, BIOS developers, and Western Digital.
The objective is to finalize the ATAPI specification around which
these companies will design and manufacture products for the personal
computer industry. Although the exact strategy has yet to be decided
upon, the document will eventually be submitted to a standards
committee for adoption.
Putting it All Together ------------------------ Support for four IDE devices Fast IDE port for disk drives Slow IDE port for CD-ROMs and tape True plug and play Lowest cost of connection Overlapped I/Os for higher performance
The Big Picture:
It is clear that the mass storage needs of the personal computer
industry are expanding to include higher performance and connectivity
requirements. Enhanced IDE was developed in response to these
requirements. The industry is already embracing Enhanced IDE and its
elements of improved functionality, performance, and connectivity by
introducing products in the calendar fourth quarter of 1993. These
products include BIOS support for >528MB IDE hard disk drives, the
shipment of >528MB IDE drives themselves, silicon and controller
products supporting fast PIO and DMA transfers, and hardware
supporting dual channel IDE for multiple device connectivity.
Momentum in the development of the industry's first standard IDE non-
disk peripherals is well underway with the industry's first IDE
CD-ROMs anticipated to ship in calendar first quarter 1994.
SCSI and IDE Scorecard:
The industry activity backed by real Enhanced IDE products means that
IDE has met the challenge in addressing the industry's new
requirements. IDE's cost effectiveness and compatibility advantages,
matched now with high performance and connectivity attributes make it
a solid storage interface solution well into the future. A new
comparison of the AT/SCSI scorecard reveals the successful approach
of Enhanced IDE:
Standard AT Interface
* The IDE interface supports two disk drives.
* IDE is a hard disk only interface.
* The IDE interface is limited to 528MB hard disk capacity as as result of the Int 13h BIOS interface used to access IDE drives.
* The IDE interface is typically limited to 2-3MB/sec host throughput.
With Enhanced IDE ----------------- * The IDE interface supports four IDE devices with dual channel IDE and more with multiple IDE connectors.
* The IDE interface supports non-disk peripherals such as IDE CD-ROM, IDE Tape.
* With LBA, the IDE interface supports up to 8.4G of hard disk capacity.
* With Mode 3 PIO and multiword DMA mode 1, data transfer rates with IDE drives can be from 11MB/sec up to 13MB/sec.
With Enhanced IDE, the IDE interface has become a mass storage
interface for personal computers and is no longer simply a disk drive
interface. Enhanced IDE complements SCSI in that it remains primarily
an internal interface solution with SCSI as an external interface
Western Digital is a registered trademark of Western Digital
Corporation. All marks mentioned herein belong to other companies.
PIO transfers are based on using the CPU to perform the data transfer
(Processor I/O) and is the standard transfer method supported within
all existing BIOS and all embedded operating system device drivers.
PIO implies compatibility with existing BIOS/OS and therefore does
not require added device driver support for operation.