Floppy-disk controller

The first floppy disk drive controller (FDC) like the first floppy disk drive (the IBM 23FD) shipped in 1971 as a component in the IBM 2385 Storage Control Unit for the IBM 2305 fixed head disk drive,[1] and of the System 370 Models 155 and 165. The IBM 3830 Storage Control Unit, a contemporaneous and quite similar controller, uses its internal processor to control a 23FD.[2] The resultant FDC is a simple implementation in IBMs’ MST hybrid circuits on a few printed circuit cards.[2] The drive, FDC and media were proprietary to IBM and although other manufacturers provided early FDDs prior to 1973 there were no standards for FDCs, drives or media.

IBM's 1973 introduction of the 3740 Data Entry System created the basic media standard for the 8-inch single sided floppy disk, IBM's "Type 1" diskette, which coupled with rapidly increasing requirements for inexpensive, removable direct access storage for many small applications caused a dramatic growth in drive and controller shipments.[3]

Prior to the introduction of special purpose integrated circuit versions, most FDCs consisted of at least one printed circuit implemented with 40 or more ICs.[4] Examples of such FDCs include:

• 1973: The FDC in IBM's 3741 is a type of microcontroller that accepts commands from the system's microprocessor ("MPU" in IBM's terminology) and executes them on the attached 33FD as independently as possible. It accepts and executes the following commands, select/stop, write check, seek lower, seek higher, read data, read I D, write data, write control, write I D, set ready, reset access counter, and nothing (no-op). It was implemented using IBM's MST hybrid circuits on the motherboard plus a separate data separator (VFO) pcb.[5] This IBM FDC did establish the IBM Type 1 diskette as the first industry standard floppy disk medium but neither its interface to the host microprocessor nor its interface to the 33FD were adopted as industry standards.
• 1974: iCOM's FD360 contained an early FDC, the CF 360, that generated industry standard media, connected to industry standard host busses, and supported industry standard FDDs.[6][7] Its FDC was implemented on a pcb approximately 12x9 inches as a state machine using 30 ICs.[8]
• 1976: Scientific Micro Systems’ FD0300 FDC[9] built on an 8-inch by 12-inch circuit board contains a microprocessor and approximately 50 integrated circuits and is designed to provide easy attachment to a number of host buses.[10]
• 1976: Shugart Associates introduced the first 5¼-inch floppy disk drive along with an associated and first FDC for this form factor, the SA4400.[11] The SA4400 performs control functions to transfer data between a host system and up to 3 disk drives using an 8-bit general purpose host interface which format disks according to a modified IBM 3740 type media format specifications. The FDC is microprocesser controlled and implemented on a 5.75 by 9.50 inch pcb with 45 ICs.[12] The drive interface and media form factors became industry standards with the media then evolving over time to support a number of different formats.
• 1977: The Apple Disc II FDC, the “Woz Machine”, is built with only 8 ICs.[4][13] It, like the much earlier IBM 3830 FDC, achieved the reduction in components thru use of the host processor and firmware. Its interface to the Apple host as well as its interface to the Apple 5¼-inch floppy disk drive is unique and it was not adopted as an industry standard.

The first FDC implemented as a special purpose integrated circuit is the Western Digital FD1771[14] announced on July 19, 1976.[15] The initial design supported a single format and required additional circuitry but over time, as a family, the design became multi-sourced and evolved to support many formats and minimize external circuitry.

The NEC µPD765, announced in 1978[16] became a quasi-industry standard when it was adopted in the original IBM PC (1981); the FDC was physically located on its own adapter card along with support circuitry. Other vendors such as Intel produced compatible parts. This design evolved over time into a family offering an almost complete FDC on a chip.[17]

In early 1987, Intel introduced the 82072 CHMOS High Integrated Floppy Disk Controller for use in industry standard PC computers.[18][19]

Ultimately in most computer systems the FDC became a part of a Super I/O chip or a Southbridge chip.[17][20] [21]

A floppy disk stores binary data not as a series of values, but a series of changes in value. Each of these changes, recorded in the polarity of the magnetic recording media, causes a voltage to be induced in the drive head as the disk surface rotates past it. It is the timing of these polarization changes and the resulting spikes of voltage that encode the ones and zeros of the original data. One of the functions of the controller is to turn the original data into the proper pattern of polarizations during writing, and then recreate it during reads.

As the storage is based on timing, and that timing is easily affected by mechanical and electrical disturbances, accurately reading the data requires some sort of reference signal, the clock. As the on-disk timing is constantly changing, the clock signal has to be provided by the disk itself. To do this, the original data is modified with extra transitions to allow the clock signal to be encoded in the data and then use clock recovery during reads to recreate the original signal. Some controllers require this encoding to be performed externally, but most designs provide standard encodings like FM and MFM.

The controller also provides a number of other services to control the drive mechanism itself. These typically include the movement of the drive head to center over the separate tracks on the disk, tracking the location of the head and returning it to zero, and sometimes functionally to format a disk based on simple inputs like the number of tracks, sectors per track and number of bytes per sector.

To produce a complete system, the controller has to be combined with additional circuitry or software that acts as a bridge between the controller and the host system. In some systems, like the Apple II and IBM PC, this is controlled by software running on the computer's host microprocessor and the drive interface is connected directly to the processor using an expansion card. On other systems, like the Commodore 64 and Atari 8-bit family, there is no direct path from the controller to the host CPU and a second processor like the MOS 6507 or Zilog Z80 is used inside the drive for this purpose.

The original Apple II controller was in the form of a plug-in card on the host computer. It could support two drives, and the drives eliminated most of the normal onboard circuitry. This allowed Apple to arrange a deal with Shugart Associates for a simplified drive that lacked most of its normal circuitry.[4] This meant that the combined cost of a single drive and controller card was roughly the same as on other systems, but a second drive could be connected for a smaller additional cost.

The IBM PC took a more conventional approach, their adaptor card could support up to four drives; on the PC direct memory access (DMA) to the drives was performed using DMA channel 2 and IRQ 6. The diagram below shows a conventional floppy disk controller which communicates with the CPU via an Industry Standard Architecture (ISA) bus or similar bus and communicates with the floppy disk drive with a 34 pin ribbon cable. An alternative arrangement that is more usual in recent designs has the FDC included in a super I/O chip which communicates via a Low Pin Count (LPC) bus.

Block diagram showing FDC communication with the CPU and the FDD.

Most of the floppy disk controller (FDC) functions are performed by the integrated circuit but some are performed by external hardware circuits. The list of functions performed by each is given below.

The FDC has three I/O ports. These are:

• Data port
• Main status register (MSR)
• Digital control port

The first two reside inside the FDC IC while the Control port is in the external hardware. The addresses of these three ports are as follows.

A controller connects to one or more drives using a flat ribbon cable, 50 wires for 8" drives and 34 wires for 3.5" & 5.25" drives. A "universal cable" has four drive connectors, two each for 3.5" & 5.25" drives.[22] In the IBM PC family and compatibles, a twist in the cable is used to distinguish disk drives by the socket to which they are connected. All drives are installed with the same drive select address set, and the twist in the cable interchanges the drive select lines at the socket. The drive that is at the far end of the cable would also have a terminating resistor installed to maintain signal quality.[23]

More detailed descriptions of the interface signals including alternative meanings are contained in manufacturer's specifications for drives or host controllers.

When the controller and disk drive are assembled as one device, as it is the case with some external floppy disk drives, e.g., Commodore 1540 and USB floppy disk drives,[24] the internal floppy disk drive and its interface are unchanged, while the assembled device presents a different interface such as IEEE-488, parallel port or USB.

Many mutually incompatible floppy disk formats are possible; aside from the physical format on the disk, incompatible file systems are also possible.

[35]

Sides:

• SS (or 1S) – Single sided
• DS (or 2S) – Double sided

Density:

• SD (or 1D) – Single density (FM)
• DD (or 2D) – Double density (most often MFM)
• QD (or 4D) – Quad density
• HD – High density
• ED – Extra-high density
• TD – Triple density

A setup disk of Japanese Microsoft Office 4.3, provided with 3.5" 1.2 MB and 1440 KB formats.

Primarily in Japan there are 3.5" high-density floppy drives that support three modes of disk formats instead of the normal two – 1440 KB (2 MB unformatted), 1.2 MB (1.6 MB unformatted) and 720 KB (1 MB unformatted). Originally, the high-density modes for 3.5" floppy drives in Japan only supported a capacity of 1.2 MB instead of the 1440 KB capacity that was used elsewhere.[36] While the more common 1440 KB format spun at 300 rpm, the 1.2 MB formats instead spun at 360 rpm, thereby closely resembling the geometries of ether the 1.2 MB format with 80 tracks, 15 sectors per track, and 512 bytes per sector previously found on 5.25" high-density floppy disks or the 1.2 MB format with 77 tracks, 8 sectors per track, and 1,024 bytes per sector previously found on 8" double-density floppy disks. Later Japanese floppy drives incorporated support for both high-density formats (as well as the double-density format), hence the name 3-mode. Some BIOSes have a configuration setting to enable this mode for floppy drives supporting it.[37]

• Western Digital FD1771
• Integrated Woz Machine (IWM)
• Paula (Amiga controller)
• Floppy disk drive interface
• List of floppy disk formats

• de Boyne Pollard, Jonathan (2003). "There is no such thing as a 3.5-inch floppy disc". Frequently Given Answers.
• ISO/IEC 8860-1:1987 Double-Density (DD)
• ISO/IEC 9529-1:1989 High-Density (HD)
• ISO 10994-1:1992 Extra-high-density (ED)
• ECMA-147
• NEC µPD72070 - Floppy Disk Controller Specification Version 2.0 (PDF). 2.0 preliminary. NEC Corporation. October 1991. Archived from the original (PDF) on 2017-03-20. Retrieved 2017-03-20.
• Shah, Katen A. (1996) [September 1992, April 1992]. Intel 82077SL for Super-Dense Floppies (PDF) (Application Note) (2 ed.). Intel Corporation, IMD Marketing. AP-358, 292093-002. Archived from the original (PDF) on 2017-06-19. Retrieved 2017-06-19.
• Johnson, Herb (2009). "The first floppy controller for CP/M and S-100?". Retrieved 2022-09-09.

• viralpatel.net A Tutorial on Programming Floppy Disk Controller
• isdaman.com Programming Floppy Disk Controllers