A look at a vintage PC's disk storage subsystem

I've been doing a lot of work recently with vintage PC disk storage subsystems. I'm hoping to present a talk on this topic at the VCF East event in April 2026. Here's a preview of some of the material I'm planning to present.

 As always, comments and questions are encouraged. 

Parts of a disk storage subsystem 

A disk storage system can be separated in several logical functions:

• A computer bus interface, usually DMA, which moves data from processor memory into the storage subsystem.
• A storage controller, which provides an interface the computer can use to address storage units, commonly 512-byte blocks.
• A data formatter, which formats the data for storage on the media. Most disks are formatted with each disk sector having a (hopefully) read-only address area, followed by a read/write data area. Each of these areas contain synchronization preambles and CRC or ECC postambles.
• A data separator, which controls how the data bits are written to the storage media and recovers the data when read. This may be frequency modulation (FM), Modified FM (MFM), or a variety of other formats. 

The ST-506 interface

The ST-506 interface put almost all of the electronics on a separate controller which plugged into a PC's backplane. Since a single controller could be connected to several disks, this saved money by allowing each drive to be simpler (or "dumb").

A common control cable was daisy-chained from drive to drive, while data passed over separate per-disk data cable in a star topology. Most such drives used MFM encoding, and data transfer was limited to about 5 MB per second

Since all of the "smarts" are on the controller, every controller was at liberty to manage the data as it wanted as long as it adhered to the basic interface. A mind-numbing matrix of synchronization preamble formats, address formats, CRC polynomials, and ECC error correction codes make moving a drive from one model controller to another nearly impossible.

The ESDI interface 

 

The ST-506 interface came with a number of serious issues. It was limited in the speed at which it could transfer data between the data separator and the drive. Seeking from one cylinder to another required significant controller involvement. And it limited the development of newer media encodings.

Disk drive manufacturers responded to the limitations of the ST-506 interface by moving the data separator from the controller onto the disk drive itself. This allowed them to design the data separator for the specific media of the individual drive.

Although the ESDI interface used the same cabling as ST-506, many of the signals have different functions. Rather than passing flux transition to the controller, the data separator on the drive recovers the data clock and data bits itself and passes this information to the controller across the data cable.

This reduced the amount of data passed between the drive and the controller, allowing an immediate doubling of data rate to 10 MB per second, with up to 24 MB per second anticipated. The "semi-intelligent" drive electronics provide the controller with information about the geometry of the drive, reducing the amount of configuration data the system software must provide to the controller.

The Integrated Drive Electronics interface 

Although ESDI was a major evolutionary step in disk drive interfaces, it was short-lived. By the time as the ANSI specification for ESDI was finalized, it had become economical to move even more of the controller functions into the drive electronics. This left only the host interface in the controller, with all other functions on the drive itself. Thus begat what we know as "IDE".  

These graphics came from a late-1980s era disk controller chip datasheet, so you can't hold me responsible for calling IDE "intelligent".

One of the major advantages of IDE was that disks formatted on one system were generally portable to other systems, assuming a common file system (such as Microsoft's FAT file system). Another is that drive manufacturers were no longer constrained to the limitations of the interface to the data separator on the drive, as the formatter was once again local to the data separator. This opened the door to innovative recording techniques that have seen huge jumps in storage capacities.