Read the original article here.

---

The Parallel Port: A Deep Dive into a Legacy Interface

In the journey of understanding how early computers worked from the ground up, studying their input/output (I/O) interfaces is crucial. These interfaces were the primary means by which the central processing unit (CPU) communicated with the outside world – peripherals like keyboards, mice, monitors, and printers. Among the most significant I/O ports in the era of personal computing was the parallel port. While largely replaced by technologies like USB and Ethernet today, the parallel port played a vital role for decades and offers valuable lessons in digital communication, hardware control, and the evolution of computer interfaces.

What is a Parallel Port?

Definition: Parallel Port A type of interface found on early computers that sends multiple bits of data simultaneously across multiple data lines, in contrast to serial ports which send data one bit at a time. It was primarily used for connecting peripherals like printers.

The fundamental difference between a parallel port and a serial port lies in how data is transmitted. A serial port sends data bits sequentially over a single data line (plus ground and control lines). A parallel port, however, uses multiple data lines (typically 8) to send an entire byte (8 bits) or more at once. This parallel transmission allows for potentially faster data transfer compared to early serial ports operating at low bit rates.

Think of it like moving furniture:

• Serial: Moving one item at a time down a narrow hallway.
• Parallel: Moving multiple items side-by-side down a wide corridor.

While parallel transmission is faster in principle over short distances due to sending more data simultaneously, it requires more wires in the cable and connector, making them bulkier than contemporary serial ports. The complexity of synchronizing multiple data lines also introduces challenges, especially over longer distances where signal timing differences (skew) become significant.

The term "parallel port" became most famously associated with the printer port, specifically the interface standardized by IEEE 1284, commonly found on IBM PC-compatible computers from the 1980s through the early 2000s.

A Historical Journey: The Evolution of the Parallel Port

Understanding the parallel port's history reveals how hardware interfaces evolve to meet changing needs and performance demands.

The Centronics Origin (Late 1970s)

The parallel port as we know it largely originated with Centronics Data Computer Corp. As they developed early, low-cost dot matrix printers, they needed an efficient way to receive character data from the host computer.

Instead of building complex circuitry in the printer to reassemble bits arriving serially, they designed an interface that accepted a full 8-bit ASCII character simultaneously on 8 data lines. This significantly simplified the printer's electronics.

Context: Dot Matrix Printing An early printing technology where characters are formed by striking a grid (matrix) of pins against an ink ribbon onto paper. The pattern of dots created by the pins forms the character shape. Early printers used a vertical row of pins (e.g., 7 or 9), and the print head would move horizontally, firing pins at specific points to build the character column by column. The speed of this mechanical process determined how quickly the printer could accept data.

To manage the flow of data, a simple handshake protocol was devised:

1. The host computer places the 8 bits of data onto the data lines.
2. The host pulls a control line called STROBE low. This signals the printer that valid data is available on the data lines.
3. The printer reads the data and pulls a status line called BUSY high. This tells the host to wait.
4. The printer processes the character (e.g., prints it or stores it in a buffer).
5. Once ready for the next character, the printer pulls BUSY low again.
6. The host detects BUSY going low and repeats the process.

Other control lines existed, such as AUTOFEED (to automatically advance paper after each line) and INITIALIZE (to reset the printer). Status lines like ERROR, SELECT (online), and PAPER OUT allowed the printer to report its state back to the host, though these were initially limited.

The Centronics Connector: Centronics used a surplus 36-pin micro ribbon connector for their printers. Although only 21 pins were initially used (8 data, several control/status, and grounds), this connector became synonymous with the interface on the printer side.

Definition: Centronics Connector A 36-pin micro ribbon connector commonly used on the printer side of the parallel port interface, originating with Centronics printers in the 1970s.

While the 36-pin connector became standard on the printer side, host manufacturers used various connectors. This led to a need for different cables depending on the computer's parallel port connector (e.g., DC-37, 25-pin card edge, 50-pin micro ribbon).

The IBM PC Implementation (1981)

When IBM introduced the IBM Personal Computer, they included a parallel printer adapter. They chose the DB25 connector for the PC side, standardizing the now-familiar parallel cable with a DB25M connector on the PC end and a 36-pin Centronics connector on the printer end.

The original IBM implementation had a limited form of bidirectionality: the 8 data lines could be used for output, and 4 status lines could be used for input. This capability was intended for peripherals other than simple printers but saw little initial use and was even removed in some later IBM PC revisions before being formally reintroduced with the IBM PS/2 line in 1987.

Introducing Bidirectionality: Bi-Tronics and IBM Byte Mode

As peripherals became more sophisticated (e.g., providing detailed status messages, requiring configuration), the limited status lines of the basic Centronics interface were insufficient. The need arose for the host to send commands to the peripheral and for the peripheral to send more than just basic status back to the host.

• Bi-Tronics (Nibble Mode): Developed by Hewlett-Packard (HP) for their LaserJet printers in 1993. This innovation used the existing 4 status input pins (ERROR, SELECT, PE, BUSY) to send data from the printer back to the host in 4-bit chunks (nibbles). Two nibble transfers could constitute an 8-bit byte. This mode was software-driven on the host and used existing hardware pins, making it relatively easy to implement.

  Definition: Nibble Mode A method for a parallel port to receive 8 bits of data from a peripheral (like a printer) by sending the data in two 4-bit chunks (nibbles) over the existing status lines (Error, Select, PE, Busy).

• IBM Byte Mode: Reintroduced more formally with the IBM PS/2. This mode allowed the main 8 data pins to be used for input as well as output, enabling the peripheral to send 8 bits (a byte) back to the host simultaneously. This required the parallel port hardware on the host side to be capable of bidirectional data transfer on the data lines.

  Definition: Byte Mode A method allowing a parallel port to receive 8 bits of data from a peripheral simultaneously using the main data pins. This requires the host port hardware to be capable of bidirectional data transfer on the data lines.

Higher Performance: EPP and ECP

The increasing demand for devices like scanners and multi-function printers (which involve scanning and printing) necessitated much higher data transfer speeds in both directions than the basic Centronics or even nibble/byte modes could provide.

• Enhanced Parallel Port (EPP): Originally developed by Zenith Electronics, EPP focused on improving the handshake protocol speed. It allowed for faster data transfer (up to ~2 MB/s) in both directions, making it suitable for devices like external storage drives (Zip drives, CD-ROMs) and scanners that needed to move large amounts of data quickly.

  Definition: Enhanced Parallel Port (EPP) An IEEE 1284 parallel port mode designed for higher-speed bidirectional communication (up to 2 MB/s), particularly suited for non-printer peripherals like scanners or storage devices. It uses a faster, hardware-driven handshake protocol than basic Centronics or IBM byte mode.

• Extended Capability Port (ECP): Developed by Microsoft and HP, ECP was designed specifically for printers and scanners. It not only offered high-speed bidirectional transfer (up to ~2.5 MB/s) but also included features to reduce the CPU load, such as Direct Memory Access (DMA) and Run-Length Encoding (RLE) compression, which is particularly useful for transferring images like faxes or black-and-white scans.

  Definition: Extended Capability Port (ECP) An IEEE 1284 parallel port mode that provides high-speed bidirectional communication (up to 2.5 MB/s) with features like DMA support and run-length encoding (RLE) compression, making it efficient for transferring data like scanned images.

  Definition: Direct Memory Access (DMA) A feature of computer systems that allows certain hardware subsystems (like parallel port controllers, disk drive controllers, or network interface cards) to access system memory independently of the central processing unit (CPU). This allows the CPU to perform other tasks while data is being transferred between memory and the peripheral, significantly improving system performance for high-throughput operations.

  Definition: Run-Length Encoding (RLE) A simple form of data compression where sequences of identical data values (runs) are stored as a single value and its count, rather than as the original sequence. ECP used RLE to compress repeating patterns often found in scanned images before transmission, reducing the amount of data that needed to be sent.

The IEEE 1284 Standard (1994)

Recognizing the proliferation of different modes and handshake variations, the Institute of Electrical and Electronics Engineers (IEEE) standardized the parallel port as IEEE 1284. This standard formally defined:

• Compatibility Mode: The original Centronics-style handshake.
• Nibble Mode: The HP Bi-Tronics method for receiving data.
• Byte Mode: The IBM method for receiving data over the data pins.
• EPP and ECP Modes: The high-speed bidirectional modes.
• Various connectors and cable types.
• A method for daisy chaining up to eight devices from a single port (though this was rarely implemented in practice for consumer devices).

Definition: IEEE 1284 An international standard that defines the characteristics of the parallel port interface, including various communication modes (Compatibility, Nibble, Byte, EPP, ECP), connectors, cables, and signaling methods. It standardized the de facto Centronics and later IBM enhancements.

The standard also refined the handshake protocols, recommending "busy-by-character" (BUSY toggles once per character received) over the older "busy-by-line" (BUSY toggles whenever any data line changes state), which reduced the number of interruptions and improved efficiency.

The Dataproducts Alternative

While the Centronics/IBM lineage dominated the personal computer market, another parallel interface standard, developed by Dataproducts, was common on mainframes and minicomputers. This used different connectors (DC-37 on the host, DD-50 or M-50 on the printer) and offered variations for longer cable runs, including a "long-line" version using differential signaling for increased noise immunity over distances up to 500 feet.

Definition: Differential Signaling A method of electrical signaling that uses two complementary signals per data line (one normal, one inverted) transmitted over two wires. The receiver looks at the difference between the two signals, making the communication more immune to noise and allowing for longer cable runs compared to single-ended signaling.

IBM PC Implementation Details: Hardware and Software

For anyone building or understanding vintage PCs, knowing the specific implementation details is key.

Connectors and Pinouts

The two most common connectors are the DB25 (female) on the PC side and the Centronics 36 (male) on the peripheral side. The pins are divided into several groups:

• Data Pins (8): Carry the actual data byte being transferred (e.g., an ASCII character). These were initially output-only in the basic Centronics mode but became bidirectional in Byte, EPP, and ECP modes.
• Control Pins (4): Driven by the host to control the peripheral's actions (STROBE, AUTOFEED, INITIALIZE, SELECT IN).
• Status Pins (5): Driven by the peripheral to report its status or signal handshake events back to the host (ACK, BUSY, SELECT, ERROR, PAPER OUT).
• Ground Pins: Provide a common electrical reference.

(Note: Detailed pin tables are often available in technical manuals or online resources like BeyondLogic.org, which go into the specific pin numbers for DB25 and Centronics 36 and their signal names.)

It's important to note that some lines are "inverted" – meaning a logic low voltage (near 0V) represents a TRUE or active state, while a logic high voltage (near +5V) represents a FALSE or inactive state. This is common in TTL/CMOS logic interfacing.

I/O Port Addresses and IRQs

In the architecture of IBM PC-compatible computers, hardware devices are accessed by the CPU through specific Input/Output (I/O) port addresses. The parallel port controller chip is mapped to a range of these addresses.

Definition: I/O Port Address A memory address used by the CPU to communicate directly with hardware devices or peripherals, such as a parallel port controller. The CPU uses special instructions (like IN and OUT in x86 assembly) to read data from or write data to these addresses.

Traditionally, IBM PCs allocated standard I/O ranges and Interrupt Request (IRQ) lines for parallel ports:

• LPT1: Typically uses base address 0x3BC and IRQ 7. This address was often associated with parallel ports found on older Monochrome Display Adapter (MDA) or Hercules graphics cards.
• LPT2: Typically uses base address 0x378 and IRQ 7. This became the most common default address for parallel ports built into motherboards or provided on dedicated I/O cards.
• LPT3: Typically uses base address 0x278 and IRQ 5. Used for additional parallel ports.

The assignment of LPT numbers (Logical Parallel Terminal) was usually handled by the system's BIOS during boot-up, based on which addresses it detected active parallel port hardware. For instance, if no port was found at 0x3BC, the port at 0x378 would be assigned as LPT1.

Definition: IRQ (Interrupt Request) A hardware signal used by peripherals to request the CPU's attention. When a device needs to perform an I/O operation or signal completion (like the printer finishing a character and becoming BUSY low, or completing a buffer), it can raise an IRQ. The CPU then pauses its current task, saves its state, and jumps to a specific routine (an Interrupt Service Routine or ISR) to handle the device's request. This allows the CPU to do other work instead of constantly checking the device's status (polling).

Sharing IRQ lines between parallel ports (or other devices) was technically possible but often problematic, potentially requiring devices to be used in less efficient polled mode rather than interrupt-driven mode.

The BIOS stores the base addresses it assigned to LPT1, LPT2, etc., in a specific memory area called the BIOS Data Area (BDA), typically starting around address 0000h:0408h. Programmers could read this location to find the assigned address for a specific LPT port.

Software Access: LPT Names and Programming

In operating systems like MS-DOS, parallel ports were exposed to software using standard device names derived from "Line Print Terminal" or "Local Print Terminal": LPT1, LPT2, LPT3. The primary port, LPT1, also had an alias PRN.

This allowed for simple command-line interaction:

• TYPE C:\AUTOEXEC.BAT > LPT1: would send the contents of the AUTOEXEC.BAT file directly to the printer connected to LPT1.
• The PRINT command in DOS provided background printing capabilities using the parallel port.

From a programming perspective, accessing the parallel port in simple operating systems like DOS or Windows 95/98 was straightforward using assembly language instructions or library functions that directly manipulated I/O ports (e.g., OUT DX, AL to write a byte in assembly, or C functions like outportb(address, value)).

However, in modern operating systems like Windows NT/2000/XP and Unix/Linux, the CPU operates in protected mode with different security rings (or privilege levels).

Definition: Security Ring (Protection Ring) A mechanism used in some computer architectures and operating systems to enforce levels of privilege. Hardware and software resources are assigned to different rings (e.g., Ring 0 for the operating system kernel, Ring 3 for user applications). Code running in a higher-numbered ring (less privileged) is typically prevented from directly accessing resources designated for a lower-numbered ring (more privileged), like directly manipulating hardware I/O ports.

User-level applications run in a less privileged ring (Ring 3). Direct access to hardware I/O ports, which are managed by the operating system kernel (Ring 0), is blocked for security and stability reasons. To access the parallel port in these OSs, programs need to use special drivers or APIs provided by the operating system or third-party libraries. This ensures that hardware access is controlled and arbitrated correctly between different applications.

In Linux, parallel ports are often accessible as device files like /dev/lp0. Programming access can involve using specific kernel modules or libraries (ppdev, libieee1284) which provide controlled access from user space after requesting appropriate permissions.

Applications Beyond Printers

While the primary association is with printers, the parallel port's ability to transfer 8 bits at a time and its various control lines made it versatile for connecting a wide range of other peripherals before USB became ubiquitous.

Examples include:

• Software Dongles: Small hardware keys connected to the port as a form of copy protection. The software would check for the presence and specific response of the dongle via the parallel port.
• External Storage: Devices like Iomega Zip drives, external CD-ROM/writers, and tape drives used parallel ports, often leveraging EPP mode for decent transfer speeds.
• Scanners: Flatbed scanners utilized the parallel port, frequently using ECP mode due to its speed, DMA, and RLE capabilities.
• Modems: Some external modems connected via the parallel port.
• Input Devices: Early gamepads and joysticks sometimes used parallel port interfaces.
• File Transfer: Utilities like MS-DOS INTERLNK and INTERSRV allowed connecting two computers via their parallel (or serial) ports to share files.
• Specialized Hardware: EPROM programmers, data acquisition cards, and custom control circuits frequently used the parallel port for direct, relatively fast digital I/O.
• SCSI Adapters: Adapters were available to connect SCSI devices (hard drives, CD-ROMs, scanners) to a parallel port, often trading speed for convenience.
• Early Digital Audio: Devices like the Covox Speech Thing or hobbyist projects like OPL2LPT and OPL3LPT used the parallel port's data and control lines to output simple digital audio or control sound chips.

Decline and Legacy

The parallel port's prominence faded with the advent of the Universal Serial Bus (USB) in the late 1990s and early 2000s. USB offered several advantages:

• Simplicity: A single, smaller connector type for many devices.
• Speed: Even early USB 1.1 (12 Mbps) offered better potential throughput than SPP (0.15 MB/s), and later USB 2.0 (480 Mbps) and 3.0+ were vastly faster than ECP/EPP.
• Hot-Plugging: Devices could be connected or disconnected without restarting the computer.
• Power: USB could provide power to peripherals, simplifying cabling.
• Daisy Chaining/Hubs: Easier connection of multiple devices.

Network printing (using Ethernet or Wi-Fi) also replaced the need for a direct physical connection to a computer for many printers.

As a result, parallel ports are considered "legacy ports" and are rarely included on modern consumer computers or laptops. The large DB25 connector also occupies significant space, a factor in shrinking device sizes.

However, the parallel port hasn't entirely disappeared:

• Legacy Hardware Support: USB-to-parallel adapters, PCI/PCIe expansion cards, and network print servers allow older parallel peripherals to still be used with modern systems.
• Hobbyist Electronics: For electronics enthusiasts prototyping circuits, the parallel port remains attractive because it offers direct access to multiple digital I/O lines from a PC, requiring relatively simple external circuitry compared to implementing a USB interface.
• Industrial/CNC Control: In some industrial automation and Computer Numerical Control (CNC) milling applications, the parallel port is still used for direct, low-level control of motors and sensors, particularly with operating systems like LinuxCNC which are designed for real-time hardware control.

Relevance to "Building a Computer from Scratch"

Studying the parallel port provides critical insights for understanding older computer architectures:

1. I/O Architecture: It demonstrates a classic model of CPU-peripheral communication via memory-mapped I/O (specifically, I/O port addresses) and interrupt handling (IRQs).
2. Hardware/Software Interaction: It highlights how software (drivers, applications) directly interacts with specific hardware addresses and how operating systems manage and protect these resources.
3. Communication Protocols: It illustrates the need for handshake signals to ensure reliable data transfer between devices operating at different speeds. The evolution from simple Centronics to EPP/ECP shows how protocols adapt for increased speed and functionality.
4. Interface Design Trade-offs: Comparing parallel with serial highlights the design choices involved in creating interfaces – trading off the number of wires, potential speed, distance, and complexity.
5. Evolution of Standards: The progression from de facto Centronics to IBM's variant to the formal IEEE 1284 standard demonstrates how successful technologies are eventually documented and unified.
6. Legacy and Compatibility: Understanding parallel ports helps explain why older software or hardware might require specific OS versions or drivers for direct hardware access, illustrating concepts like protected mode and privilege levels.

By examining the parallel port, you gain a tangible understanding of how computers moved data to the outside world in a foundational era of personal computing, providing valuable context for appreciating the interfaces we use today.