Read the original article here.

---

Expansion Cards: Adding Capabilities to Your Computer System

A Key Component in 'The Lost Art of Building a Computer from Scratch'

In the journey of understanding computer architecture from the ground up, expansion cards represent a fundamental concept: modularity. They are the building blocks that allow a basic computer system to be customized and extended to perform a vast array of tasks. Instead of needing a completely different machine for every job, expansion cards provide a standardized way to add specialized hardware capabilities.

What is an Expansion Card?

At its core, an expansion card is a piece of hardware designed to be added to a computer's main circuitry to enhance its functionality.

Expansion Card: A printed circuit board (PCB) that contains electronic components and connectors, designed to be inserted into a compatible slot on a computer's motherboard (or backplane) to add or expand features and capabilities. Synonyms include expansion board, adapter card, peripheral card, or accessory card.

These cards rely on specific physical and electrical interfaces provided by the motherboard or backplane.

Expansion Slot (or Bus Slot): An electrical connector on a motherboard or backplane specifically designed to receive an expansion card. These slots provide the physical connection points and the electrical interface (the bus) through which the card communicates with the rest of the computer system.

The interaction between the expansion card and the main system is facilitated by a bus.

Bus: In the context of expansion cards, a bus is a system of electrical pathways, connectors, and communication protocols that allows components within a computer (like the CPU, memory, and expansion cards) to exchange data, addresses, control signals, and power.

The Concept of Modularity Through Expansion

The ability to add or swap expansion cards is crucial for creating flexible and adaptable computer systems. Before integrated circuits allowed complex functions to be placed on a single chip or motherboard, even basic computer functions like the CPU, memory, and I/O controllers were often separate, large boards or modules. Expansion cards carried this modular principle forward into the era of smaller, more integrated computers, starting with minicomputers and then microcomputers.

By designing a standard slot and bus interface, computer manufacturers could create a base system (motherboard, CPU, basic memory) and rely on third-party vendors (or themselves) to create cards that added specific functions. This meant:

1. Customization: Users could configure their computer for specific needs (e.g., adding a high-quality sound card for audio production, or a specialized data acquisition card for scientific work).
2. Upgradability: As new technologies emerged (like faster graphics or network standards), users could upgrade specific components without replacing the entire computer.
3. Cost-Effectiveness: Manufacturers could produce a standardized motherboard in high volume, and specialized functions could be added later via cards, potentially reducing the cost of the base system.

In the early days of personal computing, especially when building a system "from scratch" (or assembling components), understanding which cards were needed and how they interacted with the system was paramount. The base motherboard often provided very little built-in functionality.

A Look Back: History and Evolution

The concept of modular, plug-in boards predates the modern personal computer. Early vacuum-tube computers and later minicomputers like the PDP-8 utilized modular boards for processor, memory, and I/O, communicating via a passive backplane.

Backplane: A board that contains expansion slots connected via a bus. In systems with a passive backplane, the CPU is often on one of the expansion cards plugged into the backplane, rather than being on the main board itself. This allows for extreme modularity, where almost every system component is on a separate card.

The first microcomputer to feature expansion slots was the Micral N in 1973. A significant step towards standardization in the early microcomputer era was the Altair 8800 (1974–1975), which introduced the S-100 bus. This became a multi-manufacturer standard, characteristic of many early "build-it-yourself" or kit-based computers. These often used passive backplanes where the CPU, memory, and I/O boards all plugged into slots.

Alongside these early standards, proprietary bus implementations existed, such as the unique 50-pin expansion slots of the original Apple II computer (1977).

The Rise of the PC Bus Standards

The IBM PC, introduced in 1981, established a dominant lineage of expansion buses that heavily influenced personal computing. Understanding these is key to appreciating the history of PC hardware assembly.

1. ISA (Industry Standard Architecture):

   • Initially called the PC bus (8-bit) with the IBM PC (1981) and XT (1983).
   • Extended to 16 bits with the IBM AT (1984), adding a second connector while maintaining backward compatibility for 8-bit cards.
   • Context for Building from Scratch: Early ISA cards required significant manual configuration. Users had to set jumpers or DIP switches on the card to assign system resources like:
     • I/O Port Addresses: Specific memory locations used by devices for communication.
     • Interrupt Request (IRQ) Lines: Signals devices use to get the CPU's attention.
     • DMA (Direct Memory Access) Channels: Allows devices to transfer data directly to/from memory without involving the CPU for every byte.
   • Misconfiguration of these settings could lead to conflicts, where multiple devices tried to use the same resource, causing system instability or preventing devices from working. This troubleshooting was a crucial skill in early PC assembly.
   • ISA became a de facto standard, widely adopted by clone manufacturers.

2. MCA (Micro Channel architecture):

   • IBM's proprietary replacement for ISA, introduced with the PS/2 systems in 1987.
   • Offered technical improvements like bus mastering and plug-and-play configuration (reducing the need for manual jumper setting).
   • Context: Despite technical advantages, its proprietary nature and licensing fees led to it being largely rejected by other PC manufacturers, who stuck with and further developed ISA and later standards.

3. EISA (Extended Industry Standard Architecture):

   • A 32-bit extension of ISA, developed by a consortium of PC competitors ("Gang of Nine") as an open alternative to MCA.
   • Maintained backward compatibility with 16-bit ISA cards.
   • Used on some high-end PC motherboards until the mid-1990s but was eventually superseded by PCI.

4. Local Buses (e.g., VESA Local Bus - VLB):

   • Appeared in the late 1980s/early 1990s. These were temporary solutions to provide faster access directly to the CPU's bus, primarily for performance-critical components like video cards.
   • VLB was tied closely to the Intel 80486 CPU's architecture. While faster for specific tasks, they weren't a general-purpose replacement for ISA.

5. PCI (Peripheral Component Interconnect):

   • Introduced by Intel in the early 1990s (widely adopted around 1993 with the Pentium).
   • A significant step up from ISA, offering higher speeds and support for bus mastering (allowing devices to control the bus and transfer data directly to/from memory without constant CPU intervention, improving performance).
   • Natively supported Plug-and-Play (PnP), where the system could automatically assign resources (IRQ, I/O addresses) to cards, vastly simplifying installation compared to ISA.
   • Became the dominant expansion bus, displacing both ISA and EISA, and remained standard on motherboards for many years.

6. AGP (Accelerated Graphics Port):

   • Introduced in 1997 as a dedicated high-speed interface specifically for video cards.
   • Logically connected to the PCI bus but provided a direct, high-bandwidth path to system memory for graphics textures, improving 3D performance.
   • Typically, a motherboard had only one AGP slot.

7. PCI Express (PCIe):

   • Approved in 2004, this is the dominant standard today.
   • Represents a fundamental shift from parallel bus architectures (like PCI) to a serial, point-to-point connection based on high-speed "lanes."
   • More scalable and faster than PCI, with different slot sizes (x1, x4, x8, x16) corresponding to the number of lanes.
   • Supports the logical PCI protocol but over a much faster physical layer.

Other historical PC-centric buses include PC/104 (an embedded version of ISA) and unique buses like the Tandy PLUS or IBM PCjr sidecar interfaces.

Expansion in Other Computing Families

While the PC lineage is prominent, other computer platforms also relied heavily on expansion cards:

• Apple: The Apple II used proprietary slots. Later Macintosh systems used Processor Direct Slots (PDS - direct CPU connection), NuBus (a plug-and-play standard also used in workstations), before eventually adopting PCI in the mid-1990s.
• Commodore Amiga: Used the Zorro II bus, which also featured plug-and-play capabilities.
• Industrial/Scientific Systems: Standards like HP-IB (later IEEE-488 or GPIB) for instrument control, VMEbus, STD Bus, and Sun Microsystems' SBus were used in various specialized computing environments.

Even many video game consoles featured expansion ports or cartridge slots that functionally acted as buses, allowing the addition of peripherals or memory, even if the modules weren't always in the "card" form factor.

Common Applications of Expansion Cards

What kinds of features were (and are) commonly added via expansion cards?

• Graphics Cards (Video Cards): Adding display capabilities or enhancing graphics performance. Essential in early PCs as graphics were not built-in.
• Network Interface Cards (NICs): Connecting the computer to a network (Ethernet, Wi-Fi). Allowed connectivity before it was common on motherboards.
• Sound Cards: Providing higher quality audio input/output than basic built-in sound (or adding sound capability where none existed).
• Storage Controllers: Connecting hard drives, SSDs, or other storage devices using interfaces like SATA, SAS, or older standards like IDE and ST-506.
• Modems: Enabling communication over telephone lines for dial-up internet or faxing.
• Specialized I/O Cards: Adding serial ports (RS-232), parallel ports, USB ports, or other specific interfaces needed for connecting industrial equipment, scientific instruments, or older peripherals.
• Data Acquisition Cards: For collecting data from sensors or external signals.
• Video Capture Cards: For digitizing video input from cameras or other sources.
• RAID Controllers: Managing multiple hard drives for improved performance or data redundancy.
• Wireless Adapters: Adding Wi-Fi or Bluetooth connectivity.

In systems built from scratch, selecting the appropriate expansion cards is as critical as choosing the CPU or motherboard; they define the system's capabilities.

Physical Construction and Form Factors

An expansion card is a printed circuit board with components mounted on it.

• Edge Connector/Pin Header: One edge of the card has the electrical contacts that plug directly into the expansion slot on the motherboard or backplane. The specific layout and number of contacts depend on the bus standard (ISA, PCI, PCIe, etc.).
• External Connectors: For cards that interface with external devices (like graphics, network, or audio cards), connectors (HDMI, Ethernet port, audio jacks) are mounted on a metal bracket at one end of the card. This bracket secures the card in the computer case and provides access points on the back of the machine.
• Size and Slots: Standard desktop cases and motherboards accommodate cards of a certain height and length. Modern high-performance cards (especially graphics cards) may be dual-slot or even triple-slot, occupying the space of more than one physical slot due to large heatsinks and fans needed for cooling.
• Low-Profile Cards: Smaller cards designed for compact computer cases (like Home Theater PCs or Small Form Factor systems). There are specific standards for "low profile PCI" or "low profile PCIe" which define smaller bracket and board dimensions.
• Power and Heat: As more cards are added, the system's power supply must be adequate, and cooling becomes important to dissipate heat generated by the cards.

Related Concept: Daughterboards and Riser Cards

While similar in appearance (they are both PCBs that plug into something), daughterboards often serve a slightly different role than traditional expansion cards.

Daughterboard (or Daughtercard, Mezzanine Board, Piggyback Board): A secondary circuit board that plugs into a connector on another board (like a motherboard or another expansion card), rather than directly into a standard expansion bus slot. Daughterboards often have only internal connections or add specific modular features to the host board's design.

• Purpose:

  • Internal Modularity: Adding specific features internally without needing a full expansion slot (e.g., adding a specialized network controller or a CPU voltage regulator module).
  • Form Factor Solutions: Riser cards are a specific type of daughterboard that plug into a motherboard slot and provide slots parallel to the motherboard. This allows expansion cards to be installed sideways relative to the motherboard, useful in slim or rackmount cases where vertical space is limited.
  • Adding Features to a Base Design: Manufacturers can add a daughterboard to a core product design to create variations (e.g., adding a modem or a specific interface to a standard development board).
  • Development Boards: Many modern development platforms like Arduino ("shields"), BeagleBone ("capes"), and Raspberry Pi ("HATs") use proprietary connector systems where smaller boards plug onto the main board to add sensors, interfaces, displays, etc. These are essentially modern examples of daughterboards.

• Mezzanine Boards: This term is often used for daughterboards that stack parallel to the main board, connecting via a high-density connector. They are common in embedded systems and industrial hardware (e.g., PMC, XMC, AMC standards).

Daughterboards represent another layer of modularity, allowing designers to build systems from sub-modules, sometimes adding functionality directly to the main board's features rather than accessing the system via a shared bus slot.

Key Expansion Standards

Here are some of the notable expansion bus and card standards discussed:

• Early / Historic: S-100, Apple II Slots, Acorn Tube, IBM PC/XT Bus (8-bit ISA)
• PC Lineage: ISA (16-bit), MCA, EISA, VESA Local Bus (VLB), PCI, AGP, PCI Express (PCIe), PC/104
• Laptop/External: PC Card/PCMCIA, CardBus, ExpressCard, Mini PCI/PCIe (internal)
• Other Platforms: Apple PDS, NuBus (Macintosh), Zorro II (Amiga), SBus (Sun)
• Industrial/Scientific: IEEE-488 (GPIB), VMEbus, STD Bus, FPGA Mezzanine Card (FMC), PCI Mezzanine Card (PMC)

Conclusion

Understanding expansion cards is fundamental to comprehending how computer systems gained flexibility, power, and adaptability over their history. From the early days where almost every function was on a separate board plugged into a backplane, through the era of adding graphics, sound, and network capabilities to basic motherboards, to today's high-speed PCIe cards, the concept of adding modular hardware via a standardized interface has been a constant. For anyone exploring "The Lost Art of Building a Computer from Scratch," recognizing the role of the expansion bus and the various cards that interface with it is key to appreciating the evolution of computer architecture and the choices involved in configuring a system.