'; window.popUpWin.document.write(zhtm); window.popUpWin.document.close(); // Johnny Jackson 4/28/98 } //-->

Upgrading & Repairing PCs

Floppy Disk Drives

This chapter examines in detail floppy disk drives and disks. It explores how floppy disk drives and disks function, how DOS uses a disk, what types of disk drives and disks are available, and how to properly install and service drives and disks.

Development of the Floppy Disk Drive

Alan Shugart is generally credited with inventing the floppy drive while working for IBM in the late 1960s. In 1967, he headed the disk drive development team at IBM's San Jose lab, when and where the floppy drive was created. One of Shugart's senior engineers, David Noble, actually proposed the flexible media (then 8 inches in diameter) and the protective jacket with the fabric lining. Shugart left IBM in 1969, and took more than 100 IBM engineers with him to Memorex. He was nicknamed "The Pied Piper" because of the loyalty exhibited by the many staff members who followed him. In 1973, he left Memorex, again taking with him a number of associates, and started Shugart Associates to develop and manufacture floppy drives. The floppy interface developed by Shugart is still the basis of all PC floppy drives. IBM used this interface in the PC, enabling them to use off-the-shelf third-party drives instead of custom building their own solutions.

Shugart wanted to incorporate processors and floppy drives into complete microcomputer systems at that time, but the financial backers of the new Shugart Associates wanted him to concentrate on floppy drives only. He quit (or was forced to quit) Shugart Associates in 1974, right before they introduced the mini-floppy (5 1/4-inch) diskette drive, which of course became the standard eventually used by personal computers, rapidly replacing the 8-inch drives. Shugart Associates also introduced the Shugart Associates System Interface (SASI), which was later renamed Small Computer Systems Interface (SCSI) when it was formally approved by the ANSI committee in 1986. After being forced to leave, Shugart attempted to legally force Shugart Associates to remove his name from the company, but failed. The remnants of Shugart Associates still operates today as Shugart Corporation.

For the next few years, Shugart took time off, ran a bar, and even dabbled in commercial fishing. In 1979, Finis Conner approached Shugart to create and market 5 1/4-inch hard disk drives. Together they founded Seagate Technology and by the end of 1979 had announced the ST-506 (6M unformatted, 5M formatted capacity) drive and interface. This drive is known as the father of all PC hard disk drives. Seagate then introduced the ST-412 (12M unformatted, 10M formatted capacity) drive, which was adopted by IBM for the original XT in 1983. IBM was Seagate's largest customer for many years. Today, Seagate Technology is the largest disk drive manufacturer in the world.

When you stop to think about it, Alan Shugart has had a tremendous effect on the PC industry. He (or his companies) has created the floppy, hard disk, and SCSI drive and controller interfaces still used today. All PC floppy drives are still based on (and compatible with) the original Shugart designs. The ST-506/412 interface was the de facto hard disk interface standard for many years and served as the basis for the ESDI and IDE interfaces as well. Shugart also created the SCSI interface, used in both IBM and Apple systems today.

As a side note, in the late 80s Finis Conner left Seagate and founded Conner Peripherals, originally wholly owned and funded by Compaq. Conner became Compaq's exclusive drive supplier, and gradually began selling drives to other system manufacturers as well. Compaq eventually cut Conner Peripherals free, selling off most (if not all) of their ownership of the company. In late 1996, Seagate bought Conner Peripherals, and has fully incorporated all of the Conner products into the Seagate line.

Drive Components

This section describes the components that make up a typical floppy drive and examines how these components operate together to read and write data--the physical operation of the drive. All floppy drives, regardless of type, consist of several basic common components. To properly install and service a disk drive, you must be able to identify these components and understand their function (see Figure 13.1).

Read/Write Heads

A floppy disk drive normally has two read/write heads, making the modern floppy disk drive a double-sided drive. A head exists for each side of the disk, and both heads are used for reading and writing on their respective disk sides. At one time, single-sided drives were available for PC systems (the original PC had such drives), but today single-sided drives are a fading memory (see Figure 13.2).

NOTE: Many people do not realize that the first head is the bottom one. Single-sided drives, in fact, use only the bottom head; the top head is replaced by a felt pressure pad (refer to Figure 13.2). Another bit of disk trivia is that the top head (Head 1) is not directly over the bottom head--the top head is located either four or eight tracks inward from the bottom head, depending on the drive type.

FIG. 13.1  A typical full-height disk drive.

The head mechanism is moved by a motor called a head actuator. The heads can move in and out over the surface of the disk in a straight line to position themselves over various tracks. The heads move in and out tangentially to the tracks that they record on the disk. Because the top and bottom heads are mounted on the same rack, or mechanism, they move in unison and cannot move independently of each other. The heads are made of soft ferrous (iron) compounds with electromagnetic coils. Each head is a composite design, with a read/write head centered within two tunnel-erase heads in the same physical assembly (see Figure 13.3).

The recording method is called tunnel erasure; as the track is laid down, the trailing tunnel erase heads erase the outer bands of the track, trimming it cleanly on the disk. The heads force the data to be present only within a specified narrow "tunnel" on each track. This process prevents the signal from one track from being confused with the signals from adjacent tracks. If the signal were allowed to "taper off" to each side, problems would occur. The forcibly trimmed track prevents this problem.

FIG. 13.2  Single- and double-sided drive head assemblies.

Alignment is the placement of the heads with respect to the tracks they must read and write. Head alignment can be checked only against some sort of reference-standard disk recorded by a perfectly aligned machine. These types of disks are available, and you can use one to check your drive's alignment. Unfortunately, that is not practical because one calibrated analog alignment disk costs more than three new drives today!

The two heads are spring-loaded and physically grip the disk with a small amount of pressure, which means that they are in direct contact with the disk surface while reading and writing to the disk. Because PC-compatible floppy disk drives spin at only 300 or 360 RPM, this pressure does not present an excessive friction problem. Some newer disks are specially coated with Teflon or other compounds to further reduce friction and enable the disk to slide more easily under the heads. Because of the contact between the heads and the disk, a buildup of the oxide material from the disk eventually forms on the heads. The buildup periodically can be cleaned off the heads as part of a preventive-maintenance or normal service program.

FIG. 13.3  Composite construction of a typical floppy drive head.

To read and write to the disk properly, the heads must be in direct contact with the media. Very small particles of loose oxide, dust, dirt, smoke, fingerprints, or hair can cause problems with reading and writing the disk. Disk and drive manufacturer's tests have found that a spacing as little as .000032 inches (32 millionths of an inch) between the heads and the media can cause read/write errors. You now can understand why it is important to handle disks carefully and avoid touching or contaminating the surface of the disk media in any way. The rigid jacket and protective shutter for the head access aperture on the 3 1/2-inch disks is excellent for preventing problems with media contamination. 5 1/4-inch disks do not have the same protective elements; therefore, more care must be exercised in their handling.

The Head Actuator

The head actuator is a mechanical motor device that causes the heads to move in and out over the surface of a disk. These mechanisms for floppy disk drives universally use a special kind of motor, a stepper motor, that moves in both directions in an increment called a step. This type of motor does not spin around continuously; rather, the motor turns a precise specified distance and stops. Stepper motors move in fixed increments, or detents, and must stop at a particular detent position. Stepper motors are not infinitely variable in their positioning. Each increment of motion, or a multiple thereof, defines each track on the disk. The motor can be commanded by the disk controller to position itself according to any relative increment within the range of its travel. To position the heads at track 25, for example, the motor is commanded to go to the 25th detent position.

The stepper motor usually is linked to the head rack by a coiled, split steel band. The band winds and unwinds around the spindle of the stepper motor, translating the rotary motion into linear motion. Some drives use a worm gear arrangement rather than a band. With this type, the head assembly rests on a worm gear driven directly off the stepper motor shaft. Because this arrangement is more compact, you normally find worm gear actuators on the smaller 3 1/2-inch drives.

Most stepper motors used in floppy drives can step in specific increments that relate to the track spacing on the disk. Most 48 Track Per Inch (TPI) drives have a motor that steps in increments of 3.6° (degrees). This means that each 3.6° of stepper motor rotation moves the heads from one track (or cylinder) to the next. Most 96 or 135 TPI drives have a stepper motor that moves in 1.8° increments, which is exactly half of what the 48 TPI drives use. Sometimes you see this information actually printed or stamped right on the stepper motor itself, which is useful if you are trying to figure out what type of drive you have. 5 1/4-inch 360K drives are the only 48 TPI drives available and use the 3.6° increment stepper motor. All other drive types normally use the 1.8° stepper motor. On most drives, the stepper motor is a small cylindrical object near one corner of the drive.

A stepper motor usually has a full travel time of about 1/5 of a second--about 200ms. On average, a half-stroke is 100ms, and a one-third stroke is 66ms. The timing of a one-half or one-third stroke of the head-actuator mechanism often is used to determine the reported average-access time for a disk drive. Average-access time is the normal amount of time the heads spend moving at random from one track to another.

The Spindle Motor

The spindle motor spins the disk. The normal speed of rotation is either 300 or 360 RPM, depending on the type of drive. The 5 1/4-inch high-density (HD) drive is the only drive that spins at 360 RPM; all others, including the 5 1/4-inch double-density (DD), 3 1/2-inch DD, 3 1/2-inch HD, and 3 1/2-inch extra-high density (ED) drives, spin at 300 RPM.

Most earlier drives used a mechanism on which the spindle motor physically turned the disk spindle with a belt, but all modern drives use a direct-drive system with no belts. The direct-drive systems are more reliable and less expensive to manufacture, as well as smaller in size. The earlier belt-driven systems did have more rotational torque available to turn a sticky disk because of the torque multiplication factor of the belt system. Most newer direct-drive systems use an automatic torque-compensation capability that automatically sets the disk-rotation speed to a fixed 300 or 360 RPM, and compensates with additional torque for sticky disks or less torque for slippery ones. This type of drive eliminates the need to adjust the rotational speed of the drive.

Most newer direct-drive systems use this automatic-speed feature, but many earlier systems require that you periodically adjust the speed. Looking at the spindle provides you with one clue to the type of drive you have. If the spindle contains strobe marks for 50Hz and 60Hz strobe lights (fluorescent lights), the drive probably has an adjustment for speed somewhere on the drive. Drives without the strobe marks almost always include an automatic tachometer-control circuit that eliminates the need for adjustment. The technique for setting the speed involves operating the drive under fluorescent lighting and adjusting the rotational speed until the strobe marks appear motionless, much like the "wagon wheel effect" you see in old Western movies. The procedure is described later in this chapter in the "Setting the Floppy Drive Speed Adjustment" section.

Circuit Boards

A disk drive always incorporates one or more logic boards, which are circuit boards that contain the circuitry used to control the head actuator, read/write heads, spindle motor, disk sensors, and any other components on the drive. The logic board represents the drive's interface to the controller board in the system unit.

The standard interface used by all PC types of floppy disk drives is the Shugart Associates SA-400 interface, which is based on the NEC 765 controller chip. The interface, invented by Shugart in the 1970s, has been the basis of most floppy disk interfacing. The selection of this industry-standard interface is the reason that you can purchase "off-the-shelf" drives (raw, or bare, drives) that can plug directly into your controller.

TIP: Logic boards for a drive can fail and usually are difficult to obtain as a spare part. One board often costs more than replacing the entire drive. I recommend keeping failed or misaligned drives that might otherwise be discarded so that they can be used for their remaining good parts--such as logic boards. The parts can be used to restore a failing drive very cost-effectively.

The Faceplate

The faceplate, or bezel, is the plastic piece that comprises the front of the drive. These pieces, usually removable, come in different colors and configurations.

Most drives use a bezel slightly wider than the drive. These types of drives must be installed from the front of a system because the faceplate is slightly wider than the hole in the system-unit case. Other drive faceplates are the same width as the drive's chassis; these drives can be installed from the rear--an advantage in some cases. In the later-version XT systems, for example, IBM uses this design in its drives so that two half-height drives can be bolted together as a unit and then slid in from the rear to clear the mounting-bracket and screw hardware. On occasion, I have filed the edges of a drive faceplate to install the drive from the rear of a system--which sometimes can make installation much easier.

Connectors

Nearly all disk drives have at least two connectors--one for power to run the drive, and the other to carry the control and data signals to and from the drive. These connectors are fairly standardized in the computer industry; a four-pin in-line connector (called Mate-N-Lock, by AMP), in both a large and small style is used for power (see Figure 13.4); and a 34-pin connector in both edge and pin header designs is used for the data and control signals. 5 1/4-inch drives normally use the large style power connector and the 34-pin edge type connector, whereas most 3 1/2-inch drives use the smaller version of the power connector and the 34-pin header type logic connector. The drive controller and logic connectors and pinouts are detailed later in this chapter as well as in Appen-dix A, "Vendor List."

FIG. 13.4  A disk drive female power supply cable connector.

Both the large and small power connectors from the power supply are female plugs. They plug into the male portion, which is attached to the drive itself. One common problem with upgrading an older system with 3 1/2-inch drives is that your power supply only has the large style connectors, whereas the drive has the small style. An adapter cable is available from Radio Shack (Cat. No. 278-765) and other sources that converts the large style power connector to the proper small style used on most 3 1/2-inch drives.

The following chart shows the definition of the pins on the drive power-cable con-nectors:

Large Power Connector Small Power Signal Wire Color Connector
Pin 1 Pin 4 +12 Vdc Yellow
Pin 2 Pin 3 Ground Black
Pin 3 Pin 2 Ground Black
Pin 4 Pin 1 +5 Vdc Red

NOTE: Note that the pin designations are reversed between the large- and small-style power connectors. Also, it is important to know that not all manufacturers follow the wire color coding properly. I have seen instances in which all the wires are a single color (for example, black), or the wire colors are actually reversed from normal! For example, I once purchased the Radio Shack power connector adapter cables just mentioned that had all the wire colors backwards. This was not really a problem as the adapter cable was wired correctly from end to end, but it was disconcerting to see the red wire in the power supply connector attach to a yellow wire in the adapter (and vice versa)!

Not all drives use the standard separate power and signal connectors. IBM, for example, uses either a single 34-pin or single 40-pin header connector for both power and floppy controller connections in most of the PS/2 systems. In some older PS/2 systems, for example, IBM used a special version of a Mitsubishi 3 1/2-inch 1.44M drive called the MF-355W-99, which has a single 40-pin power/signal connector. Other PS/2 systems use a Mitsubishi 3 1/2-inch 2.88M drive called the MF356C-799MA, which uses a single 34-pin header connector for both power and signal connections.

Most standard PC compatible systems use 3 1/2-inch drives with a 34-pin signal connector and a separate small style power connector. For older systems, many drive manufacturers also sell 3 1/2-inch drives installed in a 5 1/4-inch frame assembly and have a special adapter built in that allows the larger power connector and standard edge type signal connectors to be used. These drives included an adapter that enables the standard large style power connector, 34-pin edge type control, and data connector to be used. Because no cable adapters are required and they install in a 5 1/4-inch half-height bay, these types of drives are ideal for upgrading earlier systems. Most 3 1/2-inch drive- upgrade kits sold today are similar and include the drive, appropriate adapters for the power and control and data cables, a 5 1/4-inch frame adapter and faceplate, and rails for AT installations. The frame adapter and faceplate enable the drive to be installed where a 5 1/4-inch half-height drive normally would go.

Drive-Configuration Devices

Most floppy drives come properly configured for PC installation. In some cases, if the drive is used or not properly configured to begin with, you will have to check or change the configuration yourself. Most drives have a stable of jumpers and switches, and many drives are different from each other. You will find no standards for what these jumpers and switches are called, where they should be located, or how they should be implemented. There are some general guidelines to follow, but in order to set up a specific drive correctly and know all the options available, you must have information from the drive's manufacturer, normally found in the original equipment manufacturer's (OEM) manual. The manual is a "must-have" item when you purchase a disk drive.

Many drives have the following configuration settings:

Drive Select

Floppy drives are connected by a cabling arrangement called a daisy chain. The name is descriptive because the cable is strung from controller to drive to drive in a single chain. All drives have a drive select (sometimes called DS) jumper that must be set to indicate a certain drive's physical drive number. Some drives allow four settings, as that was what the original SA-400 floppy interface called for, but the controllers used in PC systems support only two drives on a single daisy-chain cable. Some controllers support four drives but only on two separate cables--each one a daisy chain with a maximum of two drives.

Every drive on a particular cable must be set to have unique drive select settings. In a normal configuration, the drive you want to respond as the first drive (A:) is set to the first drive select position, and the drive you want to respond as the second drive (B:) is set to the second drive-select position. On some drives, the usable DS jumper positions are labeled DS0 and DS1; other drives use the numbers DS1 and DS2 for the same settings. For some drives then, a setting of DS0 is drive A:. For others, however, DS1 indicates drive A:.

NOTE: If you have incorrect DS settings, both drives respond simultaneously (both lights come on at the same time) or neither drive responds at all.

The type of cable you use controls the drive select configuration. Most systems have a special twist in the floppy cable that electrically changes the DS configuration of the drive plugged in after the twist. This twist causes a drive physically set to the second DS position (B:) to appear to the controller to be set to the first DS position (A:) and vice versa. With such a cable, both drives have to be set to the same DS setting for them to work. Normally, both drives should be set to the second DS position. The drive plugged into the connector farthest from the controller, which is after the twist in the cable, then would have the physical second-DS-position setting appear to be changed to a first-DS-position setting. Then the system would see this drive as A:, and the drive plugged into the middle cable connector still would appear as B:. A typical daisy-chain drive cable with this included "twist" is connected as shown in Figure 13.5.

FIG. 13.5  A floppy controller cable showing the location of the twist in lines 10-16.

An IBM-style floppy cable is a 34-pin cable with lines 10-16 sliced out and cross-wired (twisted) between the drive connectors (refer to Figure 13.5). This twisting "cross-wires" the first and second drive-select and motor-enable signals, and therefore inverts the DS setting of the drive following the twist. All the drives in a system using this type of cable, therefore--whether you want them to be A: or B:--are physically jumpered the same way; installation and configuration are simplified because both floppies can be preset to the second DS position. Some drives used by IBM, in fact, have had the DS "jumper" setting permanently soldered into the drive logic board.

Most drives you purchase have the DS jumper already set to the second position, which is correct for the majority of systems that use a cable with the twisted lines. Although this setting is correct for the majority of systems, if you are using a cable with only a single floppy drive and no provisions for adding a second one (in other words, with only one drive connector attached, and no twist in lines 10-16), then the DS setting you make on the drive is exactly what the controller sees. You can attach only one drive, and it should appear to the system as A:--therefore, set the drive to the first DS position.

Terminating Resistors

Any signal carrying electronic media or cable with multiple connections can be thought of as an electrical bus. In almost all cases, a bus must be terminated properly at each end with terminating resistors to allow signals to travel along the bus error free. Terminating resistors are designed to absorb any signals that reach the end of a cabling system or bus so that no reflection of the signal echoes, or bounces, back down the line in the opposite direction. Engineers sometimes call this effect signal ringing. Simply put, noise and distortion can disrupt the original signal and prevent proper communications between the drive and controller. Another function of proper termination is to place the proper resistive load on the output drivers in the controller and drive.

Most older 5-1/4 inch drives use a terminating resistor in the drive plugged into the physical end of a cable. The function of this resistor is to prevent reflections or echoes of signals from reaching the end of the cable. Most removable terminating resistors used in 5-1/4 inch drives have resistance values of 150 to 330 ohms.

In a typical cabling arrangement with two 5 1/4-inch floppies, for example, the terminating resistor is installed in drive A: (at the end of the cable), and this resistor is removed from the other floppy drive on the same cable (B:). The letter to which the drive responds is not important in relation to terminator settings; the important issue is that the drive at the end of the cable has the resistor installed and functioning, and that other drives on the same cable have the resistor disabled or removed.

Most 3 1/2-inch drives have permanently installed, non-configurable terminating resistors. This is the best possible setup because you never have to remove or install them, and there are never any TR jumpers to set. Although some call this automatic termination, technically the 3 1/2-inch drives use a technique called distributed termination. With distributed termination, each 3 1/2-inch drive has a much higher value (1,000 to 1,500 ohm) terminating resistor permanently installed, and therefore carries a part of the termination load. These terminating resistors are fixed permanently to the drive and never have to be removed or adjusted.

When you mix 5 1/4-inch and 3 1/2-inch drives, you should enable or disable the terminators on the 5 1/4-inch drives appropriately, according to their position on the cable, and ignore the non-changeable settings on the 3 1/2-inch drives.

A terminating resistor usually looks like a memory chip--a 16-pin dual inline package (DIP) device. The device is actually a group of eight resistors physically wired in parallel with each other to terminate separately each of the eight data lines in the interface subsystem. Normally, this "chip" is a different color from other black chips on the drive. Orange, yellow, blue, or white are common colors for a terminating resistor. Some drives use a resistor network in a single inline pin (SIP) package, which looks like a slender device with eight or more pins in a line. IBM always labels the resistor with a T-RES sticker for easy identification on their drives. On some systems, the resistor is a built-in device enabled or disabled by a jumper or series of switches (often labeled TM or TR).

CAUTION: Be aware that not all drives use the same type of terminating resistor, however, and it might be physically located in different places on different manufacturer's drive models. The OEM manual for the drive comes in handy in this situation because it shows the location, physical appearance, enabling and disabling instructions, and even the precise value required for the resistors.

Do not lose the terminator if you remove it from a drive; you might need to reinstall it later if you relocate the drive to a different position in a system or even to a different system.

Figure 13.6 shows the location and appearance of the terminating resistor or switches on a typical floppy drive. Because most 3 1/2-inch drives have a form of automatic termination, there is no termination to configure.

You don't have to worry about the controller end of the cable because a terminating resistor network is built into the controller to properly terminate that end of the bus.

Note that in many cases, even if the termination is improper a system seems to work fine, although the likelihood of read and write errors may be increased. In older systems with only 5-1/2 inch drives, the drives do not work properly at all unless termination is properly configured.

Diskette Changeline

The standard PC floppy controller and drive use a special signal on pin 34 called Diskette Changeline to determine whether the disk has been changed, or more accurately, whether the same disk loaded during the previous disk access is still in the drive. Disk Change is a pulsed signal that changes a status register in the controller to let the system know that a disk has been either inserted or ejected. This register is set to indicate that a disk has been inserted or removed (changed) by default. The register is cleared when the controller sends a step pulse to the drive and the drive responds, acknowledging that the heads have moved. At this point, the system knows that a specific disk is in the drive. If the disk change signal is not received before the next access, the system can assume that the same disk is still in the drive. Any information read into memory during the previous access can therefore be reused without rereading the disk.

FIG. 13.6  A typical floppy drive terminating resistor, or termination switches.

Because of this process, systems can buffer or cache the contents of the file allocation table (FAT) or directory structure of a disk in the system's memory. By eliminating unnecessary rereads of these areas of the disk, the apparent speed of the drive is increased. If you move the door lever or eject button on a drive that supports the disk change signal, the DC pulse is sent to the controller, thus resetting the register and indicating that the disk has been changed. This procedure causes the system to purge buffered or cached data that had been read from the disk because the system then cannot be sure that the same disk is still in the drive.

AT-class systems use the DC signal to increase significantly the speed of the floppy interface. Because the AT can detect whether you have changed the disk, the AT can keep a copy of the disk's directory and FAT information in RAM buffers. On every subsequent disk access, the operations are much faster because the information does not have to be reread from the disk in every individual access. If the DC signal has been reset (has a value of 1), the AT knows that the disk has been changed and appropriately rereads the information from the disk.

You can observe the effects of the DC signal by trying a simple experiment. Boot DOS on an AT-class system and place a formatted floppy disk with data on it in drive A:. Drive A: can be any type of drive except 5 1/4-inch double-density, although the disk you use can be anything the drive can read, including a double-density 360K disk, if you want. Then type the following command: DIR A: The disk drive lights up, and the directory is displayed. Note the amount of time spent reading the disk before the directory is displayed on-screen. Without touching the drive, enter the DIR A: command again, and watch the drive-access light and screen. Note again the amount of time that passes before the directory is displayed. The drive A: directory should appear almost instantly the second time because virtually no time is spent actually reading the disk. The directory information was simply read back from RAM buffers or cache rather than read again from the disk. Now eject and re-insert the disk. Type the DIR A: command again. The disk again takes some time reading the directory before displaying anything because the system "thinks" that you changed the disk.

Older PC and XT low-density controllers (and systems) are not affected by the status of the DC signal. These systems "don't care" about signals on pin 34. The PC and XT systems always operate under the assumption that the disk is changed before every access, and they reread the disk directory and FAT each time--one reason why these systems are slower in using the floppy disk drives.

One interesting problem can occur when certain drives are installed in a 16-bit or greater system. As mentioned, some drives use pin 34 for a "Ready" (RDY) signal. The RDY signal is sent whenever a disk is installed and rotating in the drive. If you install a drive that has pin 34 set to send RDY, the system "thinks" that it is continuously receiving a disk change signal, which causes problems. Usually the drive fails with a Drive not ready error and is inoperable. The only reason that the RDY signal exists on some drives is that it happens to be a part of the standard Shugart SA-400 disk interface; however, it has never been used in PC systems.

The biggest problem occurs if the drive is not sending the DC signal on pin 34, and it should. If a system is told (through CMOS setup) that the drive is any other type than a 360K (which cannot ever send the DC signal), the system expects the drive to send DC whenever a disk has been ejected. If the drive is not configured properly to send the signal, the system never recognizes that a disk has been changed. Therefore, even if you do change the disk, the AT still acts as though the first disk is in the drive and holds the first disk's directory and FAT information in RAM. This can be dangerous because the FAT and directory information from the first disk can be partially written to any subsequent disks written to in the drive.

CAUTION: If you ever have seen an AT-class system with a floppy drive that shows "phantom directories" of the previously installed disk, even after you have changed or removed it, you have experienced this problem firsthand. The negative side effect is that all disks after the first one you place in this system are in extreme danger. You likely will overwrite the directories and FATs of many disks with information from the first disk.

If even possible at all, data recovery from such a catastrophe can require quite a bit of work with utility programs such as Norton Utilities. These problems with Disk Change most often are traced to an incorrectly configured drive. This problem will be covered in more detail in the section "Phantom Directory (Disk Change) Problems" later in this chapter.

If the drive is a 5 1/4-inch 360K drive, set the status of pin 34 to Open (disconnected) regardless of the type of system in which you are installing the drive. The only other option normally found for pin 34 on 360K drives is RDY, which is incorrect. If you are using only a low-density controller, as in a PC or XT, pin 34 is ignored no matter what is sent on it.

If the drive you are installing is a 5 1/4-inch 1.2M or 3 1/2-inch 720K, 1.44M, or 2.88M drive, be sure to set pin 34 to send the Disk Change (DC) signal. The basic rule is simple: For 360K drives only, pin 34 = Open (disconnected) For any other drive, pin 34 = Disk Change

Media Sensor

This configuration item exists only on the 3 1/2-inch 1.44M or 2.88M drives. The jumper selection, called the media sensor (MS) jumper, must be set to enable a special media sensor in the disk drive, which senses a media sensor hole found only in the 1.44M HD and the 2.88M ED floppy disks. The labeling of this jumper (or jumpers) varies greatly between different drives. In many drives, the jumpers are permanently set (enabled) and cannot be changed.

The three types of configurations with regards to media sensing are as follows:

Most systems use a passive media sensor arrangement. The passive media sensor setup enables the drive to determine the level of recording strength to use and is required for most installations of these drives, because of a bug in the design of the Western Digital hard disk and floppy controllers used by IBM in the AT systems. This bug prevents the controller from properly instructing the drive to switch to double-density mode when you write or format DD disks. With the media sensor enabled, the drive no longer depends on the controller for density mode switching and relies only on the drive's media sensor. Unless you are sure that your disk controller does not have this flaw, make sure that your HD drive includes a media sensor (some older or manufacturer-specific drives do not), and that it is properly enabled.

The 2.88M drives universally rely on media sensors to determine the proper mode of operation. The 2.88M drives, in fact, have two separate media sensors because the ED disks include a media sensor hole in a different position than the HD disks.

With only a few exceptions, HD 3 1/2-inch drives installed in most PC-compatible systems do not operate properly in double-density mode unless the drive has control over the write current (recording level) via an installed and enabled media sensor. Exceptions are found primarily in systems with floppy controllers integrated on the motherboard, including most older IBM PS/2 and Compaq systems as well as most laptop or notebook systems from other manufacturers. These systems have floppy controllers without the bug referred to earlier, and can correctly switch the mode of the drive without the aid of the media sensor.

In these systems, it technically does not matter whether you enable the media sensor. If the media sensor is enabled, the drive mode is controlled by the disk you insert, as is the case with most PC-compatible systems. If the media sensor is not enabled, the drive mode is controlled by the floppy controller, which in turn is controlled by DOS.

If a disk is already formatted correctly, DOS reads the volume boot sector to determine the current disk format, and the controller then switches the drive to the appropriate mode. If the disk has not been formatted yet, DOS has no idea what type of disk it is, and the drive remains in its native HD or ED mode.

When you format a disk in systems without an enabled media sensor (such as most PS/2s), the mode of the drive depends entirely on the FORMAT command issued by the user, regardless of the type of disk inserted. For example, if you insert a DD disk into an HD drive in an IBM PS/2 Model 70 and format the disk by entering FORMAT A:, the disk is formatted as though it is an HD disk because you did not issue the correct parameters (/F:720) to cause the FORMAT command to specify a DD format. On a system with the media sensor enabled, this type of incorrect format fails, and you see the Invalid media or Track 0 bad error message from FORMAT. In this case, the media sensor prevents an incorrect format from occurring on the disk, a safety feature most older IBM PS/2 systems lack.

Most of the newer PS/2 systems--including all those that come standard with the 2.88M drives--have what is called an active or intelligent media sensor setup. This means that the sensor not only detects what type of disk is in the drive and changes modes appropriately, but also the drive informs the controller (and the BIOS) about what type of disk is in the drive. Systems with an intelligent media sensor do not need to use the disk type parameters in the FORMAT command. In these systems, the FORMAT command automatically "knows" what type of disk is in the drive and formats it properly. With an intelligent media sensor, you never have to know what the correct format parameters are for a particular type of disk; the system figures it out for you automatically. Many high-end systems such as the newer PS/2 systems as well as high-end Hewlett-Packard PCs have this type of intelligent media sensor arrangement.

The Floppy Disk Controller

The floppy disk controller consists of the circuitry either on a separate adapter card or integrated on the motherboard, which acts as the interface between the floppy drives and the system. Most PC- and XT-class systems use a separate controller card that occupied a slot in the system. The AT systems normally have the floppy controller and hard disk controller built into the same adapter card and also plugged into a slot. In most of the more modern systems built since then, the controller is integrated on the motherboard. In any case, the electrical interface to the drives has remained largely static, with only a few exceptions.

The original IBM PC and XT system floppy controller was a 3/4-length card that could drive as many as four floppy disk drives. Two drives could be connected to a cable plugged into a 34-pin edge connector on the card, and two more drives could be plugged into a cable connected to the 37-pin connector on the bracket of this card. Figures 13.7 and 13.8 show these connectors and the pinouts for the controller.

FIG. 13.7  A PC and XT floppy controller internal connector.

The AT used a board made by Western Digital, which included both the floppy and hard disk controllers in a single adapter. The connector location and pinout for the floppy controller portion of this card is shown in Figure 13.9.

IBM used two variations of this controller during the life of the AT system. The first one was a full 4.8 inches high, which used all the vertical height possible in the AT case. This board was a variation of the Western Digital WD1002-WA2 controller sold through distributors and dealers. The second-generation card was only 4.2 inches high, which enabled it to fit into the shorter case of the XT-286 as well as the taller AT cases. This card was equivalent to the Western Digital WD1003-WA2, also sold on the open market.

FIG. 13.8  A PC and XT floppy controller external connector.

Disk Physical Specifications and Operation

PC-compatible systems now use one of as many as five standard types of floppy drives. Also, five types of disks can be used in the drives. This section examines the physical specifications and operations of these drives and disks.

Drives and disks are divided into two classes: 5 1/4-inch and 3 1/2-inch. The physical dimensions and components of a typical 5 1/4-inch disk and a 3 1/2-inch disk are shown later in this chapter.

FIG. 13.9  An AT floppy controller connector.

The physical operation of a disk drive is fairly simple to describe. The disk rotates in the drive at either 300 or 360 RPM. Most drives spin at 300 RPM; only the 5 1/4-inch 1.2M drives spin at 360 RPM (even when reading or writing 360K disks). With the disk spinning, the heads can move in and out approximately 1 inch and write either 40 or 80 tracks. The tracks are written on both sides of the disk and therefore sometimes are called cylinders. A single cylinder comprises the tracks on the top and bottom of the disk. The heads record by using a tunnel-erase procedure in which a track is written to a specified width, and then the edges of the track are erased to prevent interference with any adjacent tracks.

The tracks are recorded at different widths for different drives. Table 13.1 shows the track widths in both millimeters and inches for the five types of floppy drives supported in PC systems.

Table 13.1  Floppy Drive Track-Width Specifications

Drive Type No. of Tracks Track Width
5 1/4-inch 360K 40 per side 0.300 mm 0.0118 in.
5 1/4-inch 1.2M 80 per side 0.155 mm 0.0061 in.
3 1/2-inch 720K 80 per side 0.115 mm 0.0045 in.
3 1/2-inch 1.44M 80 per side 0.115 mm 0.0045 in.
3 1/2-inch 2.88M 80 per side 0.115 mm 0.0045 in.

The differences in recorded track width can result in data-exchange problems between 5 1/4-inch drives. The 5 1/4-inch drives are affected because the DD drives record a track width nearly twice that of the HD drives. A problem occurs, therefore, if an HD drive is used to update a DD disk with previously recorded data on it.

Even in 360K mode, the HD drive cannot completely overwrite the track left by an actual 360K drive. A problem occurs when the disk is returned to the person with the 360K drive: That drive reads the new data as embedded within the remains of the previously written track. Because the drive cannot distinguish either signal, an Abort, Retry, Ignore error message appears on-screen. The problem does not occur if a new disk (one that never has had data recorded on it) is first formatted in a 1.2M drive with the /4 option, which formats the disk as a 360K disk.

NOTE: You also can format a brand new 360K disk in a 1.2M drive with the /N:9, /T:40, or /F:360 options, depending on the DOS version. The 1.2M drive can then be used to fill the brand new and newly formatted 360K disk to its capacity, and every file will be readable on the 40-track, 360K drive.


NOTE: I use this technique all the time to exchange data disks between AT systems that have only the 1.2M drive and XT or PC systems that have only the 360K drive. The key is to start with either a new disk or one wiped clean magnetically by a bulk eraser or degaussing tool. Just reformatting the disk does not work by itself because formatting does not actually erase a disk; instead it records data across the entire disk.

Disk Magnetic Properties

A subtle problem with the way a disk drive works magnetically is that the recording volume varies depending on the type of format you are trying to apply to a disk. The high-density formats use special disks that require a much higher volume level for the recording than do the DD disks. My classes nearly always are either stumped or incorrect (unless they have read ahead in the book) when they try to answer this question: "Which type of disk is magnetically more sensitive: a 1.2M disk or a 360K disk?" If you answer that the 1.2M disk is more sensitive, you are wrong! The HD disks are approximately half as sensitive magnetically as the DD disks.

The HD disks are called high-coercivity disks also because they require a magnetic field strength much higher than do the DD disks. Magnetic field strength is measured in oersteds. The 360K floppy disks require only a 300-oersted field strength to record, and the HD 1.2M disks require a 600-oersted field strength. Because the HD disks need double the magnetic field strength for recording, you should not attempt to format a 1.2M HD disk as though it were a 360K disk, or a 360K disk as though it were a 1.2M HD disk.

An interesting problem results from this type of improper formatting: You can imprint the 360K disk magnetically with an image that is difficult to remove. The HD format places on the disk a recording at twice the strength it should be. How do you remove this recording and correct the problem? If you attempt to reformat the disk in a 360K drive, the drive writes in a reduced write-current mode and in some cases cannot overwrite the higher-volume recorded image you mistakenly placed on the disk. If you attempt to reformat the disk in the high-density drive with the /4 (or equivalent) parameter, which indicates 360K mode, the HD drive uses a reduced write-current setting and again cannot overwrite the recording.

You can correct the problem in several ways. You can throw away the disk and write it off as a learning experience, or you can use a bulk eraser or degaussing tool to demagnetize the disk. These devices can randomize all the magnetic domains on a disk and return it to an essentially factory-new condition. You can purchase a bulk-erasing device at electronic supply stores for about $25.

The opposite problem with disk formatting is not as common, but some have tried it anyway: formatting an HD disk with a DD format. You should not (and normally cannot) format a 1.2M HD disk to a 360K capacity. If you attempt to use one, the drive changes to reduced write-current mode and does not create a magnetic field strong enough to record on the "insensitive" 1.2M disk. The result in this case is normally an immediate error message from the FORMAT command: Invalid media or Track 0 bad - disk unusable. Fortunately, the system usually does not allow this particular mistake to be made.

The 3 1/2-inch drives don't have the same problems as the 5 1/4-inch drives--at least for data interchange. Because both the HD and DD drives write the same number of tracks and these tracks are always the same width, no problem occurs when one type of drive is used to overwrite data written by another type of drive. A system manufacturer therefore doesn't need to offer a DD version of the 3 1/2-inch drive for systems equipped with the HD or ED drive. The HD and ED drives can perfectly emulate the operations of the 720K DD drive, and the ED drive can perfectly emulate the 1.44M HD drive.

The HD and ED drives can be trouble, however, for inexperienced users who try to format disks to incorrect capacities. Although an ED drive can read, write, and format DD, HD, and ED disks, a disk should be formatted and written at only its specified capacity. An ED disk therefore should be formatted only to 2.88M, and never to 1.44M or 720K. You must always use a disk at its designated format capacity. You are asking for serious problems if you place a 720K disk in the A: drive of a PS/2 Model 50, 60, 70, or 80 and enter FORMAT A:. This step causes a 1.44M format to be written on the 720K disk, which renders it unreliable at best and requires a bulk eraser to reformat it correctly. If you decide to use the resulting incorrectly formatted disk, you will eventually have massive data loss.

This particular problem could have been averted if IBM had used media sensor drives in all PS/2 systems. Drives that use the disk media-sensor hole to control the drive mode are prevented from incorrectly formatting a disk. The hardware causes the FORMAT command to fail with an appropriate error message if you attempt to format the disk to an incorrect capacity.

Logical Operation

Each type of drive can create disks with different numbers of sectors and tracks. This section examines how DOS sees a drive. It gives definitions of the drives according to DOS and the definitions of cylinders and clusters.

How the Operating System Uses a Disk

To the operating system, data on your PC disks is organized in tracks and sectors. Tracks are narrow, concentric circles on a disk. Sectors are pie-shaped slices of the disk. DOS versions 1.0 and 1.1 read and write 5 1/4-inch DD disks with 40 tracks (numbered 0-39) per side and eight sectors (numbered 1-8) per track. DOS versions 2.0 and higher automatically increase the track density from eight to nine sectors for greater capacity on the same disk. On an AT with a 1.2M disk drive, DOS V3.0 supports HD 5 1/4-inch drives that format 15 sectors per track and 80 tracks per side; DOS V3.2 supports 3 1/2-inch drives that format nine sectors per track and 80 tracks per side; DOS V3.3 supports 3 1/2-inch drives that format 18 sectors per track and 80 tracks per side. The distance between tracks and, therefore, the number of tracks on a disk, is a built-in mechanical and electronic function of the drive.

Tables 13.2 and 13.3 summarize the standard disk formats supported by DOS version 5.0 and higher.

Table 13.2  5 1/4-inch Floppy Disk Drive Formats

5 1/4-Inch Floppy Disks Double Density
360K (DD)		 High Density
1.2M (HD)
Bytes per Sector 512 512
Sectors per Track 9 15
Tracks per Side 40 80
Sides 2 2
Capacity (K) 360 1,200
Capacity (Megabytes) 0.352 1.172
Capacity (Million bytes) 0.369 1.229

Table 13.3  3 1/2-inch Floppy Disk Drive Formats

3 1/2-Inch Floppy Disks Double
Density
720K (DD)		 High
Density
1.44M (HD)		 Extra-High
Density
2.88M (ED)
Bytes per Sector 512 512 512
Sectors per Track 9 18 36
Tracks per Side 80 80 80
Sides 2 2 2
Capacity (K) 720 1,440 2,880
Capacity (Megabytes) 0.703 1.406 2.813
Capacity (Million bytes) 0.737 1.475 2.949

You can calculate the capacity differences between different formats by multiplying the sectors per track by the number of tracks per side together with the constants of two sides and 512 bytes per sector.

Note that the disk capacity can actually be expressed in different ways. The most common method is to refer to the capacity of a floppy by the number of kilobytes (1,024 bytes equals 1K). This works fine for 360K and 720K disks, but is strange when applied to the 1.44M and 2.88M disks. As you can see, a 1.44M disk is really 1,440K, and not actually 1.44 megabytes. Because a megabyte is 1,024K, what we call a 1.44M disk is actually 1.406M in capacity.

Another way of expressing disk capacity is in millions of bytes. In that case, the 1.44M disk has 1.475 million bytes of capacity. To add to the confusion over capacity expression, both megabyte and millions of bytes are abbreviated as MB or M. No universally accepted standard for the definition of M or MB exists, so throughout this book I use M.

Like blank sheets of paper, new disks contain no information. Formatting a disk is similar to adding lines to the paper so that you can write straight across. Formatting places on the disk the information DOS needs to maintain a directory and file table of contents. Using the /S (system) option in the FORMAT command resembles making the paper a title page. FORMAT places on the disk the portions of DOS required to boot the system.

The operating system reserves the track nearest to the outside edge of a disk (track 0) almost entirely for its purposes. Track 0, Sector 1 contains the DOS Boot Record (DBR), or Boot Sector, the system needs to begin operation. The next few sectors contain the FATs, which act as the disk "room reservation clerk" that keeps records of which clusters or allocation units (rooms) on the disk have file information and which are empty. Finally, the next few sectors contain the root directory, in which DOS stores information about the names and starting locations of the files on the disk; you see most of this information when you use the DIR command.

In computer-industry jargon, this process is "transparent to the user," which means that you don't have to (and generally cannot) decide where information is stored on disks. That this process is "transparent," however, doesn't necessarily mean that you shouldn't be aware of the decisions DOS makes for you.

When DOS writes data, it always begins by attempting to use the earliest available data sectors on the disk. Because the file might be larger than the particular block of available sectors selected, DOS then writes the remainder of the file in the next available block of free sectors. In this manner, files can become fragmented as they are written to fill a hole on the disk created by the deletion of some smaller file. The larger file completely fills the hole; then DOS continues to look for more free space across the disk, from the outermost tracks to the innermost tracks. The rest of the file is deposited in the next available free space.

This procedure continues until eventually all the files on your disk are intertwined. This situation is not really a problem for DOS because it was designed to manage files in this way. The problem is a physical one: Retrieving a fragmented file that occupies 50 or 100 separate places across the disk takes much longer than if the file were in one piece. Also, if the files were in one piece, recovering data in the case of a disaster would be much easier. Consider unfragmenting a disk periodically simply because it can make recovery from a disk disaster much easier; many people, however, unfragment disks for the performance benefit in loading and saving files that are in one piece.

How do you unfragment a disk? DOS 6.0 and higher versions include a command called DEFRAG. This utility is actually a limited version of the Norton Utilities Speedisk program. It does not have some of the options of the more powerful Norton version and is not as fast, but it does work well in most cases. Earlier versions of DOS do not provide any easy method for unfragmenting a disk, although by backing up and restoring files, you can accomplish the goal. To unfragment a floppy disk for example, you can copy all the files one by one to an empty disk, delete the original files from the first disk, and then recopy the files. With a hard disk, you can back up all the files, reformat the disk, and restore the files. This procedure is time-consuming, to say the least.

Windows 95 also includes a Disk Defragmenter utility that not only works under the Windows graphical environment, but also operates in the background while other applications are running.

Because DOS versions earlier than 6.0 did not provide a good way to unfragment a disk, many software companies have produced utility programs that can easily unfragment disks in a clean and efficient manner. These programs can restore file contiguity without reformat and restore operations. My favorite for an extremely safe, easy, and fast un-fragmenting program is the Vopt utility by Golden Bow. In my opinion, no other unfragmenting utility even comes close to this amazing $50 package. Golden Bow's address and phone number are in Appendix B, "Glossary."

If you are using Windows 95 with long file names, be aware that many of the older defragmenter programs do not preserve these file name entries, which can cause many problems. Contact the manufacturer of any disk defragmenting utilities to ensure that they are safe to use on a Windows 95 formatted disk. Many of these programs will require updated versions to work properly.

CAUTION: Before using an unfragmenting program, make sure that you have a good backup. What shape do you think your disk would be in if the power failed during an unfragmenting session? Also, some programs have bugs or are incompatible with new releases of DOS or Windows.

Cylinders

The term cylinder usually is used in place of track. A cylinder is all the tracks under read/write heads on a drive at one time. For floppy drives, because a disk cannot have more than two sides and the drive has two heads, normally there are two tracks per cylinder. Hard disks can have many disk platters, each with two (or more) heads, for many tracks per single cylinder.

Clusters or Allocation Units

A cluster also is called an allocation unit in DOS version 4.0 and higher. The term is appropriate because a single cluster is the smallest unit of the disk that DOS can allocate when it writes a file. A cluster or allocation unit consists of one or more sectors--usually two or more. Having more than one sector per cluster reduces the FAT size and enables DOS to run faster because it has fewer individual allocation units of the disk with which to work. The tradeoff is in some wasted disk space. Because DOS can manage space only in the cluster size unit, every file consumes space on the disk in increments of one cluster.

Table 13.4 lists the default cluster sizes used by DOS for different floppy disk formats. Chapter 14, "Hard Disk Drives," discusses hard disk cluster or allocation unit sizes.

Table 13.4  DOS Default Cluster and Allocation Unit Sizes

Floppy Disk
Capacity		 Cluster/Allocation FAT Type Unit Size
5 1/4-inch, 360K 2 sectors 1,024 bytes 12-bit
5 1/4-inch, 1.2M 1 sector 512 bytes 12-bit
3 1/2-inch, 720K 2 sectors 1,024 bytes 12-bit
3 1/2-inch, 1.44M 1 sector 512 bytes 12-bit
3 1/2-inch, 2.88M 2 sectors 1,024 bytes 12-bit

K = 1,024 bytes
M = 1,048,576 bytes

The HD disks normally have smaller cluster sizes, which seems strange because these disks have many more individual sectors than do DD disks. The probable reason is that because these HD disks are faster than their DD counterparts, IBM and Microsoft thought that the decrease in wasted disk space cluster size and speed would be welcome. You learn later that the cluster size on hard disks can vary much more between different versions of DOS/Windows and different disk sizes.

Types of Floppy Drives

Five types of standard floppy drives are available for a PC-compatible system. The drives can be summarized most easily by their formatting specifications (refer to Tables 13.2 and 13.3).

Most drive types can format multiple types of disks. For example, the 3 1/2-inch ED drive can format and write on any 3 1/2-inch disk. The 5 1/4-inch HD drive also can format and write on any 5 1/4-inch disk (although, as mentioned, sometimes track-width problems occur). This drive can even create some older obsolete formats, including single-sided disks and disks with eight sectors per track.

As you can see from Table 13.5, the different disk capacities are determined by several parameters, some of which seem to remain constant on all drives, whereas others change from drive to drive. For example, all drives use 512-byte physical sectors, which remains true for hard disks as well. Note, however, that DOS treats the sector size as though it could be a changeable parameter, although the BIOS does not.

NOTE: Note also that now all standard floppy drives are double-sided. IBM has not shipped PC systems with single-sided drives since 1982; these drives are definitely considered obsolete. Also, IBM has never used any form of single-sided 3 1/2-inch drives, although that type of drive appeared in the first Apple Macintosh systems in 1984. IBM officially began selling and supporting 3 1/2-inch drives in 1986 and has used only double-sided versions of these drives.

Table 13.5  Floppy Disk Logical DOS-Format Parameters

Current Formats

Obsolete Formats

Disk Size (in.) 3 1/2" 3 1/2" 3 1/2" 5 1/4" 5 1/4" 5 1/4" 5 1/4" 5 1/4"
Disk Capacity (K) 2,880 1,440 720 1,200 360 320 180 160
Media Descriptor Byte F0h F0h F9h F9h FDh FFh FCh FEh
Sides (Heads) 2 2 2 2 2 2 1 1
Tracks per Side 80 80 80 80 40 40 40 40
Sectors per Track 36 18 9 15 9 8 9 8
Bytes per Sector 512 512 512 512 512 512 512 512
Sectors per Cluster 2 1 2 1 2 2 1 1
FAT Length (Sectors) 9 9 3 7 2 1 2 1
Number of FATs 2 2 2 2 2 2 2 2
Root Dir. Length (Sectors) 15 14 7 14 7 7 4 4
Maximum Root Entries 240 224 112 224 112 112 64 64
Total Sectors per Disk 5,760 2,880 1,440 2,400 720 640 360 320
Total Available Sectors 5,726 2,847 1,426 2,371 708 630 351 313
Total Available Clusters 2,863 2,847 713 2,371 354 315 351 313

The 360K 5 1/4-Inch Drive

The 5 1/4-inch low-density drive is designed to create a standard-format disk with 360K capacity. Although I persistently call these low-density drives, the industry term is double-density. I use low-density because I find the term double-density to be somewhat misleading, especially when I am trying to define these drives in juxtaposition to the high-density drives.

The term double-density arose from the use of the term single density to indicate a type of drive that used frequency modulation (FM) encoding to store approximately 90K on a disk. This type of obsolete drive never was used in any PC-compatible systems, but was used in some older systems such as the original Osborne-1 portable computer. When drive manufacturers changed the drives to use Modified Frequency Modulation (MFM) encoding, they began using the term double-density to indicate it, as well as the (approximately doubled) increase in recording capacity realized from this encoding method. All modern floppy disk drives use MFM encoding, including all types listed in this section. Encoding methods such as FM, MFM, and RLL (Run Length Limited) variants are discussed in Chapter 14, "Hard Disk Drives."

The 360K 5 1/4-inch drives spin at 300 RPM, which equals exactly five revolutions per second, or 200 ms per revolution. All standard floppy controllers support a 1:1 interleave, in which each sector on a specific track is numbered (and read) consecutively. To read and write to a disk at full speed, a controller sends data at a rate of 250,000 bps. Because all low-density controllers can support this data rate, virtually any controller supports this type of drive, depending on ROM BIOS code that supports these drives.

All standard PC-compatible systems include ROM BIOS support for these drives; therefore, you usually do not need special software or driver programs to use them. This statement might exclude some aftermarket (non-IBM) 360K drives for PS/2 systems that might require some type of driver in order to work. The IBM-offered units use the built-in ROM support to enable these drives to work. The only requirement usually is to run the Setup program for the machine to enable it to properly recognize these drives.

The 1.2M 5 1/4-Inch Drive

The 1.2M high-density floppy drive first appeared in the IBM AT system introduced in August 1984. The drive required the use of a new type of disk to achieve the 1.2M format capacity, but it still could read and write (although not always reliably) the lower-density 360K disks.

The 1.2M 5 1/4-inch drive normally recorded 80 cylinders of two tracks each, starting with cylinder 0, at the outside of the disk. This situation differs from the low-density 5 1/4-inch drive in its capability to record twice as many cylinders in approximately the same space on the disk. This capability alone suggests that the recording capacity for a disk would double, but that is not all. Each track normally is recorded with 15 sectors of 512 bytes each, increasing the storage capacity even more. In fact, these drives store nearly four times the data of the 360K disks. The density increase for each track required the use of special disks with a modified media designed to handle this type of recording. Because these disks initially were expensive and difficult to obtain, many users attempted incorrectly to use the low-density disks in the 1.2M 5 1/4-inch drives and format them to the higher 1.2M-density format, which results in data loss and unnecessary data-recovery operations.

A compatibility problem with the 360K drives stems from the 1.2M drive's capability to write twice as many cylinders in the same space as the 360K drives. The 1.2M drives position their heads over the same 40 cylinder positions used by the 360K drives through double stepping, a procedure in which the heads are moved every two cylinders to arrive at the correct positions for reading and writing the 40 cylinders on the 360K disks. The problem is that because the 1.2M drive normally has to write 80 cylinders in the same space in which the 360K drive writes 40, the heads of the 1.2M units had to be made dimensionally smaller. These narrow heads can have problems overwriting tracks produced by a 360K drive that has a wider head because the narrower heads on the 1.2M drive cannot "cover" the entire track area written by the 360K drive.

The 1.2M 5 1/4-inch drives spin at 360 RPM, or six revolutions per second, or 166.67ms per revolution. The drives spin at this rate no matter what type of disk is inserted--either low- or high-density. To send or receive 15 sectors (plus required overhead) six times per second, a controller must use a data-transmission rate of 500,000 bps (500KHz). All standard high- and low-density controllers support this data rate and, therefore, these drives.

This support of course depends also on proper ROM BIOS support of the controller in this mode of operation. When a standard 360K disk is running in an HD drive, it also is spinning at 360 RPM; a data rate of 300,000 bps (300KHz) therefore is required in order to work properly. All standard AT-style low- and high-density controllers support the 250KHz, 300KHz, and 500KHz data rates. The 300KHz rate is used only for HD 5 1/4-inch drives reading or writing to low-density 5 1/4-inch disks.

Virtually all standard AT-style systems have a ROM BIOS that supports the controller's operation of the 1.2M drive, including the 300KHz data rate.

The 720K 3 1/2-Inch Drive

The 720K, 3 1/2-inch, DD drives first appeared in an IBM system with the IBM Convertible laptop system introduced in 1986. In fact, all IBM systems introduced since that time have 3 1/2-inch drives as the standard supplied drives. This type of drive also is offered by IBM as an internal or external drive for the AT or XT systems.

NOTE: Outside the PC-compatible world, other computer-system vendors (Apple, Hewlett-Packard, and so on) offered 3 1/2-inch drives for their systems well before the PC-compatible world "caught on."

The 720K, 3 1/2-inch, DD drive normally records 80 cylinders of two tracks each, with nine sectors per track, resulting in the formatted capacity of 720K.

It is interesting to note that many disk manufacturers label these disks as 1.0M disks, which is true. The difference between the actual 1.0M of capacity and the usable 720K after formatting is that some space on each track is occupied by the header and trailer of each sector, the inter-sector gaps, and the index gap at the start of each track before the first sector. These spaces are not usable for data storage, and account for the differences between the unformatted and formatted capacities. Most manufacturers report the unformatted capacities because they do not know on which type of system you will format the disk. Apple Macintosh systems, for example, can store 800K of data on the same disk because of a different formatting technique.

Note also that the 720K of usable space does not account for the disk areas DOS reserves for managing the disk (boot sectors, FATs, directories, and so on) and that because of these areas, only 713K remains for file data storage.

PC-compatible systems have used 720K, 3 1/2-inch, DD drives primarily in XT-class systems because the drives operate from any low-density controller. The drives spin at 300 RPM, and therefore require only a 250KHz data rate from the controller to operate properly. This data rate is the same as the 360K disk drives, which means that any controller that supports a 360K drive also supports the 720K drives.

The only issue to consider in installing a 720K, 3 1/2-inch drive is whether the ROM BIOS offers the necessary support. An IBM system with a ROM BIOS date of 06/10/85 or later has built-in support for 720K drives and requires no driver in order to use them. If your system has an earlier ROM BIOS date, the DRIVER.SYS program from DOS V3.2 or higher--as well as the DRIVPARM CONFIG.SYS command in some OEM DOS versions--is all you need to provide the necessary software support to operate these drives. Of course, a ROM BIOS upgrade to a later version negates the need for "funny" driver software and is usually the preferred option when you add one of these drives to an older system.

The 1.44M 3 1/2-Inch Drive

The 1.44M, 3 1/2-inch, HD drives first appeared from IBM in the PS/2 product line introduced in 1987. Although IBM has not officially offered this type of drive for any of its older systems, most compatible vendors started offering the drives as options in systems immediately after IBM introduced the PS/2 system.

The drives record 80 cylinders consisting of two tracks each with 18 sectors per track, resulting in the formatted capacity of 1.44M. Most disk manufacturers label these disks as 2.0M disks, and the difference between this unformatted capacity and the formatted usable result is lost during the format. Note that the 1,440K of total formatted capacity does not account for the areas DOS reserves for file management, leaving only 1423.5K of actual file-storage area.

These drives spin at 300 RPM, and in fact must spin at that speed to operate properly with your existing high- and low-density controllers. To use the 500KHz data rate, the maximum from most standard high- and low-density floppy controllers, these drives could spin at only 300 RPM. If the drives spun at the faster 360 RPM rate of the 5 1/4-inch drives, they would have to reduce the total number of sectors per track to 15, or else the controller could not keep up. In short, the 1.44M 3 1/2-inch drives store 1.2 times the data of the 5 1/4-inch 1.2M drives, and the 1.2M drives spin exactly 1.2 times faster than the 1.44M drives. The data rates used by both HD drives are identical and compatible with the same controllers. In fact, because these 3 1/2-inch HD drives can run at the 500KHz data rate, a controller that can support a 1.2M 5 1/4-inch drive can support the 1.44M drives also. If you are using a low-density disk in the 3 1/2-inch HD drive, the data rate is reduced to 250KHz, and the disk capacity is 720K.

The primary issue in a particular system using a 1.44M 3 1/2-inch drive is one of ROM BIOS support. An IBM system with a ROM BIOS date of 11/15/85 or later has built-in support for these drives, and no external driver support program is needed. You might need a generic AT setup program because IBM's Setup program doesn't offer the 1.44M drive as an option. Another problem relates to the controller and the way it signals the HD drive to write to a low-density disk. The problem is discussed in detail in the following section.

The 2.88M 3 1/2-Inch Drive

The new 2.88M drive was developed by Toshiba Corporation in the 1980s, and was officially announced in 1987. Toshiba began production manufacturing of the drives and disks in 1989, and then several vendors began selling the drives as upgrades for systems. IBM officially adopted these drives in the PS/2 systems in 1991, and virtually all PS/2s sold since then have these drives as standard equipment. Because a 2.88M drive can fully read and write 1.44M and 720K disks, the change was an easy one. DOS version 5.0 or higher is required to support the 2.88M drives.

A number of manufacturers are making these drives, including Toshiba, Mitsubishi, Sony, and Panasonic. Unfortunately due to high media costs, these drives have not caught on, although virtually all systems today have built-in support for them.

The 2.88M ED drive uses a technique called vertical recording to achieve its great linear density of 36 sectors per track. This technique increases density by magnetizing the domains perpendicular to the recording surface. By essentially placing the magnetic domains on their ends and stacking them side-by-side, density increases enormously.

The technology for producing heads that can perform a vertical or perpendicular recording has been around a while. It is not the heads or even the drives that represent the major breakthrough in technology; rather, it is the media that is special. Standard disks have magnetic particles shaped like tiny needles that lie on the surface of the disk. Orienting these acicular particles in a perpendicular manner to enable vertical recording is very difficult. The particles on a barium-ferrite floppy disk are shaped like tiny, flat, hexagonal platelets that easily can be arranged to have their axes of magnetization perpendicular to the plane of recording. Although barium ferrite has been used as a material in the construction of permanent magnets, no one has been able to reduce the grain size of the platelets enough for HD recordings.

Toshiba has perfected a glass-crystallization process for manufacturing the ultra-fine platelets used in coating the barium-ferrite disks. This technology, patented by Toshiba, is being licensed to a number of disk manufacturers, all of whom are producing barium-ferrite disks using Toshiba's process. Toshiba also made certain modifications to the design of standard disk drive heads to enable them to read and write the new barium-ferrite disks, as well as standard cobalt or ferrite disks. This technology is being used not only in floppy drives but also is appearing in a variety of tape drive formats.

The disks are called 4M disks in reference to their unformatted capacity. Actual formatted capacity is 2,880K, or 2.88M. Because of space lost in the formatting process, as well as space occupied by the volume boot sector, FATs, and root directory, the total usable storage space is 2,863K.

To support the 2.88M drive, modifications to the disk controller circuitry were required because these drives spin at the same 300 RPM but have an astonishing 36 sectors per track. Because all floppy disks are formatted with consecutively numbered sectors (1:1 interleave), these 36 sectors have to be read and written in the same time it takes a 1.44M drive to read and write 18 sectors. This requires that the controller support a much higher data transmission rate of 1MHz (1 million bps). Most of the older floppy controllers either found on an adapter card or built into the motherboard support only the maximum of 500KHz data rate used by the 1.44M drives. To upgrade to 2.88M drives would require that the controller be changed to one that supports the higher 1MHz data rate.

An additional support issue is the ROM BIOS. The BIOS must have support for the controller and the capability to specify and accept the 2.88M drive as a CMOS setting. Newer motherboard BIOS sets from companies like Phoenix, AMI, and Award have support for the new ED controllers.

In addition to the newer IBM PS/2 systems, most newer IBM clone and compatible systems now have built-in floppy controllers and ROM BIOS software that fully supports the 2.88M drives. Adding or upgrading to a 2.88M drive in these systems is as easy as plugging in the drive and running the CMOS Setup program. For those systems that do not have this built-in support, this type of upgrade is much more difficult. Several companies offer new controllers and BIOS upgrades as well as the 2.88M drives specifically for upgrading older systems.

Although the 2.88M drives themselves are not much more expensive than the 1.44M drives they replace, the disk media is currently still very expensive. Although you can purchase 1.44M disks for around (or under) 50 cents each, the 2.88M disks can cost more than $2 per disk! This drive never really took off because it didn't help the problem much; even 2.88M is not enough for some Word docs these days.

Handling Recording Problems with 1.44M 3 1/2-Inch Drives

A serious problem awaits many users who use the 1.44M 3 1/2-inch drives: If the drive is installed improperly, any write or format operations performed incorrectly on 720K disks can end up trashing data on low-density disks. The problem is caused by the controller's incapability to signal the HD drive that a low-density recording will take place.

HD disks require a higher write-current or signal strength when they record than do the low-density disks. A low-density drive can record at only the lower write-current, which is correct for the low-density disks; the HD drive, however, needs to record at both high and low write-currents depending on which type of disk is inserted in the drive. If a signal is not sent to the HD drive telling it to lower or reduce the write-current level, the drive stays in its normal high write-current default mode, even when it records on a low-density disk. The signal normally should be sent to the drive by the controller, but many controllers do not provide this signal properly for the 1.44M drives.

The Western Digital controller used by IBM enables the reduced write-current (RWC) signal only if the controller also is sending data at the 300KHz data rate, indicating the special case of a low-density disk in a HD drive. The RWC signal is required to tell the HD drive to lower the head-writing signal strength to be proper for the low-density disks. If the signal is not sent, the drive defaults to the higher write-current, which should be used for only HD disks. If the controller is transmitting the 250KHz data rate, the controller knows that the drive must be a low-density drive and therefore no RWC signal is necessary because the low-density drives can write only with reduced current.

This situation presented a serious problem for owners of 1.44M drives using 720K disks because the drives spin the disks at 300 RPM, and in writing to a low-density disk use the 250KHz data rate--not the 300KHz rate. This setup "fools" the controller into thinking that it is sending data to a low-density drive, which causes the controller to fail to send the required RWC signal. Without the RWC signal, the drive records improperly on the disk, possibly trashing any data being written or any data already present. Because virtually all compatibles use controllers based on the design of the IBM AT floppy disk controller, most share the same problem as the IBM AT.