Hardware History: Cases, Power Supplies, & Cooling: The Early Years
Foundations of Computation : Part 24 of 24
The motherboards and firmware that connect a computer’s components and prepare the hardware to operate are only part of what makes a functioning system possible. Processors, memory, and circuit boards also need a physical structure to hold them together, a dependable source of electrical power, and a way to remove the heat they produce. Computers have always needed these three things, although the equipment used to provide them has changed considerably.
Early computers often occupied large cabinets in specially prepared rooms, with substantial electrical and cooling requirements. As electronic components became smaller and more efficient, computers could fit into smaller enclosures and operate in ordinary workplaces and homes. Size was only one consideration. Mainframes, minicomputers, supercomputers, and early personal computers each presented different challenges for their designers.
The arrival of microcomputers also changed who could build and maintain a computer. Electronics enthusiasts could construct machines from plans, assemble commercial kits, or purchase partially completed systems. Ready-built computers appeared alongside these approaches, and eventually standardized components made it easier to assemble, repair, and upgrade a computer without building its electronic circuits. A suitable case, reliable power supply, and adequate cooling helped make those machines practical.
This article follows the development of computer enclosures, electrical power, and cooling from the early decades of electronic computing through the 1990s. We will look at the engineering behind familiar desktop cases, specialized machines such as the Cray-1, and the standards that allowed computer builders to use parts from different manufacturers. By the end of the century, most computer buyers could purchase a complete system, while building and upgrading computers had become a hobby and specialty of its own.
The 1950s and 1960s: Computers as Industrial Equipment
Before computers became familiar objects on desks and in homes, they were substantial pieces of industrial equipment. Their cabinets held thousands of electronic components, they required substantial electrical infrastructure, and the heat they generated demanded careful attention. Cabinets had to protect the circuitry while giving technicians room to inspect and repair it. Electrical distribution and cooling were just as much a part of the installation as the computing equipment itself.
Mainframes: Cabinets, Power, and Cooling
Early electronic computers were often assembled into multiple cabinets or racks, with separate sections for processing, memory, power supplies, and peripheral connections. The IBM 701 and, later, the System/360 family used cabinet arrangements suited to their components and peripheral equipment. Technicians needed access to internal components for inspection, maintenance, and repairs, so cabinet design had to accommodate both the electronics and the people servicing them.
Providing electrical power was a considerable engineering challenge. Early vacuum-tube computers consumed substantial amounts of electricity, much of which ultimately became heat. Power supplies converted incoming electricity into the voltages required by the machine, while regulation helped protect sensitive circuits from fluctuations. As transistors replaced vacuum tubes, computers generally became more compact and energy-efficient, but reliable power distribution remained essential. Large installations could require dedicated electrical infrastructure, carefully planned wiring, and equipment to condition or regulate incoming power.
Removing that heat was part of keeping the machine reliable. Vacuum tubes ran hot, and densely packed electronic components could fail or behave unpredictably when temperatures rose too far. Fans moved air through cabinets, while computer rooms often required dedicated air conditioning to maintain suitable operating conditions. Some systems used liquids for more specialized thermal control. The IBM 7030 Stretch, for example, relied primarily on blower-driven air cooling for its equipment, while its magnetic-core memory used a temperature-controlled oil bath. Maintaining stable temperatures could matter as much as removing heat.
Minicomputers: Smaller Systems, Familiar Challenges
During the 1960s, minicomputers brought computing into laboratories, engineering departments, universities, and smaller organizations that could not justify a large mainframe installation. Systems such as Digital Equipment Corporation’s PDP-8 were physically smaller and generally less demanding than contemporary mainframes. However, smaller did not necessarily mean self-contained in the modern desktop sense. Many minicomputers still occupied equipment racks or cabinets alongside their storage devices, interfaces, and other peripherals.
Rack-mounted construction offered practical advantages. Components could be organized into replaceable modules, with backplanes providing electrical connections between circuit boards. Power supplies could be housed within the same cabinet, while fans and ventilation openings directed air across the electronics. A technician could replace or add a board without rebuilding the machine or opening several separate enclosures.
The rack itself was not new to computing. Standardized equipment racks had long been used in telecommunications and other technical fields. Computer manufacturers adopted similar approaches because they provided a practical way to organize electrical equipment, distribute power, manage wiring, and maintain access to components. That familiar rack arrangement still serves the same purpose in modern server rooms.
A useful distinction when examining early computer photographs is that the computer, its operator console, and its storage equipment may occupy separate cabinets. The enclosure is often part of a larger installation, not the whole machine.
Terminals and the Equipment Around the Computer
A computer installation involved more than the central machine. Operators needed consoles, displays, printers, and other equipment to interact with it. Early systems often relied on switches, indicator lamps, and printed output, while teleprinters and video terminals became increasingly important as computing developed.
Cathode-ray tube displays introduced their own enclosure and power requirements. A CRT needed high-voltage circuitry, physical protection for its glass tube, and adequate ventilation for its internal electronics. Terminals were often separate devices with their own power supplies, rather than displays drawing power directly from the computer.
Peripherals affected the layout of the entire installation. Cables connected equipment across desks, cabinets, and sometimes entire rooms. Keeping those connections accessible and protected was part of maintaining a reliable system, even before internal cable management became a familiar concern for personal-computer builders.
By the end of the 1960s, several familiar practices were already established: modular cabinets, replaceable circuit boards, regulated power supplies, forced-air cooling, and equipment designed for maintenance. Computers were becoming smaller, but the need to house, power, and cool their electronics had not changed. The arrival of affordable microprocessors in the following decade would bring those same engineering questions to a very different audience.
The 1970s: Electronics Hobbyists and the Emerging Microcomputer
While mainframes and minicomputers continued to serve businesses, universities, and research institutions, a different kind of computing began taking shape during the 1970s. The development of affordable microprocessors made it possible to build useful computers with far fewer components than earlier systems required. Electronics enthusiasts could now build a computer without the resources of a university or large company.
The early microcomputer market drew heavily from an existing culture of amateur radio operators, electronics experimenters, and people accustomed to building equipment from published plans. Some assembled circuits from individual components, while others purchased kits or partially completed machines. Ready-built computers appeared alongside kits and homebuilt machines; none of these approaches immediately replaced the others. The differences were often visible in the enclosure, power supply, and cooling arrangements—or in the absence of those features.
From Electronics Projects to Computer Kits
Long before microcomputers became practical, electronics enthusiasts were building radios, amplifiers, test equipment, and other devices from schematics published in magazines and technical books. Constructing equipment required selecting components, soldering connections, mounting circuit boards, and providing a suitable enclosure and power supply. Those skills transferred naturally to early computer projects.
Heathkit had spent years selling equipment that customers assembled themselves. The company became particularly well known for amateur radio equipment, but its catalog also included test instruments, audio equipment, televisions, and other electronic devices. Customers received components and detailed assembly instructions, allowing them to construct equipment without designing every circuit themselves. For many customers, learning how the equipment worked was part of the reason to buy a kit.
Early microcomputers followed similar approaches. The MITS Altair 8800, introduced in 1975, was offered as a kit or assembled system. Its metal enclosure contained a power supply, backplane, and circuit boards, while switches and indicator lights on the front panel allowed users to enter instructions and examine the machine’s operation. The IMSAI 8080, introduced later that year, offered a comparable approach with its own distinctive front panel.
An Altair or IMSAI looked more like a piece of electronic test equipment than a modern personal computer. Front panels provided direct access to the machine, while the internal arrangement emphasized circuit boards, expansion, and electrical connections. The case protected the electronics, but it also had to accommodate the practical needs of someone assembling, modifying, or troubleshooting the system.
Heathkit entered the computer market with the H8 in 1977. Rather than copying the Altair’s rows of toggle switches, the H8 used an octal keypad and LED display for front-panel interaction. Its enclosure, power supply, and internal expansion arrangements were designed around the company’s established kit-building model. Heathkit supplied more than a box of components: it provided a design that the purchaser could assemble, test, and maintain.
Bare Boards, Enclosures, and Power Supplies
Not every early microcomputer came with a complete enclosure. Some were sold as bare circuit boards, while others required buyers to supply their own power supplies, keyboards, displays, or storage devices. This gave hobbyists considerable freedom, but it also placed responsibility for electrical and mechanical details on the owner.
The Apple I, introduced in 1976, is a good example. Apple sold an assembled circuit board rather than a complete computer in a finished case. Buyers needed to provide a power supply, keyboard, display, and other equipment. Some owners mounted their systems in homemade wooden enclosures, while others used simpler arrangements. The familiar wooden Apple I cases seen today reflect individual constructions rather than a standard factory enclosure.
Power supplies were especially important in these arrangements. Electronic circuits required specific DC voltages, and supplying the wrong voltage or reversing a connection could damage components. Builders needed to account for transformers, rectification, filtering, voltage regulation, and adequate current capacity. They also had to consider electrical insulation, grounding, and the safe placement of components connected to household electricity.
Cooling requirements varied. Many early microprocessors consumed relatively little power compared with later desktop processors, but voltage regulators, power-supply components, memory, and other circuitry still generated heat. Metal cases, ventilation openings, and natural convection could provide sufficient cooling for some systems, while others used fans. The necessary approach depended on the complete machine, not simply the processor.
Even experienced electronics builders could save considerable work by purchasing a kit with a properly designed enclosure and power supply. A complete kit reduced the amount of mechanical and electrical design the hobbyist had to undertake independently.
When restoring a kit computer or a bare-board system, identify the required supply voltages and connector polarity before applying power. A connector that fits does not establish that a power supply is electrically compatible.
Complete Microcomputers Arrive
Not everyone interested in computers wanted to assemble one. Manufacturers quickly recognized a market for computers that arrived with their essential hardware already assembled.
The Apple II, introduced in 1977, was a much more complete product than the Apple I. Its molded plastic enclosure incorporated a keyboard, motherboard, expansion slots, and internal power supply. Buyers still needed a display and appropriate storage or peripherals, but they no longer had to construct the basic computer around an exposed circuit board.
The Apple II also used a switching power supply developed by Rod Holt. Rather than relying on a bulky conventional linear design, it converted incoming power at a higher switching frequency, allowing smaller magnetic components while providing the regulated voltages the computer required. Switching power supplies predated the Apple II, but this design illustrates how power conversion helped manufacturers fit a complete computer into a compact enclosure.
The Commodore PET, also introduced in 1977, took a different approach. Its enclosure integrated the keyboard, display, and computer electronics into a single unit. This reduced the number of separate components a purchaser needed to arrange and connect, although it also meant the enclosure had to accommodate the display’s electrical and thermal requirements.
The TRS-80 Model I, introduced by Radio Shack in the same year, offered another arrangement: a keyboard unit containing the computer electronics, accompanied by a separate monitor and power supply. By 1977, buyers could choose among several quite different ways of packaging a personal computer.
None of these approaches immediately displaced kits or homebuilt systems. Electronics enthusiasts continued constructing and modifying computers, while customers who primarily wanted to use a computer increasingly had ready-built alternatives.
Supercomputers: The Cray-1
While hobbyists were assembling microcomputers, Seymour Cray and his team were addressing a very different set of requirements. The Cray-1, introduced in 1976, was designed for scientific and technical workloads that demanded far more performance than contemporary microcomputers could provide.
The Cray-1’s distinctive C-shaped arrangement was not simply an exercise in appearance. Its densely packed electronic modules were arranged to help keep electrical connections short, an important consideration when signals needed to travel between components at very high speeds. The resulting structure became one of the most recognizable designs in computing history.
Around the base sat the machine’s famous padded bench. The seating concealed power-supply equipment and portions of the cooling infrastructure, giving the Cray-1 an appearance unlike conventional computer cabinets. It was an unusual combination of engineering and industrial design: a supercomputer whose supporting infrastructure became part of its visual identity.
Cooling the Cray-1 required considerably more than ordinary cabinet ventilation. Its densely packed circuitry produced substantial heat, which was carried away through a liquid-refrigerant cooling system. Heat passed from the electronic modules into metal cold bars containing refrigerant tubing, rather than relying on room air alone. The system illustrates how high-performance computers could require specialized thermal engineering long before liquid cooling became familiar to personal-computer enthusiasts.
The Cray-1 offers a useful comparison with the microcomputers appearing at about the same time. Hobbyists were learning to house and power computers assembled from relatively inexpensive components, while supercomputer engineers were designing entire physical systems around the electrical and thermal demands of specialized hardware. Both faced the same fundamental requirements, but the scale and solutions were dramatically different.
By the end of the 1970s, computer enclosures ranged from exposed circuit boards and homemade wooden boxes to integrated consumer machines and elaborate supercomputer installations. Power supplies and cooling systems were becoming increasingly tailored to each machine’s intended use. During the following decade, the growing personal-computer market would bring these design decisions into millions of homes and offices.
The 1980s: Personal Computers Find Their Form
During the 1980s, personal computers moved beyond the electronics hobbyist market and became increasingly common in homes, schools, and workplaces. Manufacturers experimented with different physical designs, from compact keyboard-based computers to expandable desktop systems and transportable machines. The enclosure was becoming an important part of a computer’s identity, while power supplies and cooling systems had to accommodate increasingly capable hardware.
No single enclosure design dominated every part of the market. Commodore, Atari, Sinclair, Apple, IBM, and other manufacturers approached computer construction differently. Some emphasized affordability and compactness, while others prioritized expansion, serviceability, or portability. At the same time, workstations and supercomputers continued developing specialized solutions for demanding computing environments.
Home Computers: Compact and Self-Contained
Many popular home computers of the early 1980s placed their main electronics directly beneath the keyboard. The Commodore 64, introduced in 1982, followed this approach, as did machines from Sinclair, Atari, and other manufacturers. Their plastic enclosures provided an inexpensive way to protect the electronics while keeping the computer small enough for a household desk.
These machines often used external power supplies, sometimes called power bricks. Moving the power supply outside the computer reduced the space required inside the enclosure and separated some of the power-conversion circuitry from the main electronics. However, the external supply remained an essential part of the system, and its reliability could directly affect the computer.
The Commodore 64 shows why the external supply mattered. Its original external power supply provided regulated 5-volt DC and a separate 9-volt AC output required by the computer. Failures in some of these supplies could expose the computer to excessive voltage, potentially damaging internal components. This remains a concern for people maintaining original machines today.
Cooling was generally straightforward. Many home computers relied on ventilation openings and natural convection rather than dedicated fans. Their processors consumed relatively little power, although other chips and voltage regulators could become quite warm. Enclosure design therefore needed to provide sufficient airflow without adding unnecessary cost, noise, or mechanical complexity.
The IBM PC: Standardizing the Desktop
The IBM Personal Computer, introduced in 1981, helped establish a different physical arrangement. Rather than integrating the computer beneath the keyboard, IBM used a separate horizontal metal enclosure designed to sit on a desk, often beneath the monitor.
Inside the case were the motherboard, expansion cards, power supply, and storage devices. A removable cover provided access to the internal components, while standardized expansion slots allowed owners to install additional hardware. The arrangement made the computer comparatively straightforward to service and upgrade.
The IBM PC’s power supply was mounted inside the enclosure and included a cooling fan. The fan exhausted warm air from the case as well as the supply, drawing replacement air through the computer. The power supply therefore served two purposes: providing regulated electrical power and contributing to the machine’s cooling.
The IBM PC/XT and PC/AT continued developing this general arrangement. The AT, introduced in 1984, brought changes to the motherboard and power-supply connections that influenced compatible systems for years afterward. The AT-style power supply used a mechanical switch controlling incoming AC power, rather than the software-controlled power arrangements familiar on modern computers.
The internal wiring of these machines also reflected their period. Wide ribbon cables connected floppy drives and hard drives to their controllers, while separate power connectors supplied electricity to storage devices. Expansion cards occupied substantial space, and arranging cables around them was primarily a matter of making everything fit while preserving access and adequate ventilation.
Features such as front-panel key locks and, on many compatible systems, Turbo buttons became familiar elements of desktop cases. Key locks could restrict certain forms of keyboard access, while Turbo controls on later PC compatibles often changed the machine’s operating speed to improve compatibility with software sensitive to processor timing. These features reflected the practical concerns of the period rather than purely decorative choices.
The Cray-1 and Cray-2 are good reminders that liquid cooling is not a single technology. The Cray-1 transferred heat through refrigerated metal structures; the Cray-2 immersed its modules in a dielectric liquid.
Macintosh, Amiga, and Alternative Designs
IBM-compatible computers were not the only approach to personal-computer construction. Apple’s original Macintosh, introduced in 1984, placed the computer electronics, CRT display, and floppy drive inside a compact upright enclosure. Unlike the IBM PC, the original Macintosh did not use an internal cooling fan.
The fanless design reduced operating noise but required careful consideration of heat dissipation. Ventilation openings and natural convection allowed warm air to escape, while the compact enclosure limited the available space for internal expansion and maintenance. The Macintosh demonstrated the advantages and compromises of integrating a complete computer and display into one housing.
Apple’s Macintosh II, introduced in 1987, took a different direction. Its separate desktop enclosure provided expansion slots, internal storage options, and active cooling. This made it more suitable for users who needed additional hardware and more flexible configurations.
Commodore’s Amiga family also demonstrated several approaches to enclosure design. The Amiga 1000 used a separate desktop system unit, while the Amiga 500 integrated its electronics beneath the keyboard. The Amiga 2000 offered a larger expandable desktop enclosure with internal drive bays and expansion capabilities.
Atari’s ST family similarly included compact keyboard-based machines, while later models provided more elaborate configurations. Across these competing platforms, manufacturers balanced cost, physical size, expandability, electrical requirements, and heat management in different ways.
The choice affected what an owner could do with the machine. A compact home computer could be convenient and inexpensive, but an expandable desktop system offered greater flexibility for adding storage, memory, and specialized hardware. Neither arrangement was universally superior; each reflected the intended market and capabilities of the machine.
An original Commodore 64 power supply deserves testing before use. Some aging supplies can fail in ways that send excessive voltage into the computer, so an apparently intact power brick is not necessarily a safe one.
Transportable Computers and the Beginnings of Portability
As personal computers became more useful, manufacturers began addressing another problem: moving them between locations. Early transportable computers were not particularly small or light, but they combined components that would otherwise require separate equipment.
The Osborne 1, introduced in 1981, incorporated a small CRT display, keyboard, floppy drives, and computer electronics into a transportable enclosure. Its construction required careful organization of the internal components, and the machine still depended on external AC power.
The Compaq Portable, introduced in 1983, brought a similar concept to the IBM-compatible market. Its suitcase-like enclosure protected the display and electronics during transport, while the keyboard could be attached over the front of the machine. Internal power distribution and ventilation had to work within a confined space that was considerably different from a conventional desktop case.
Making a computer portable required more than a smaller circuit board. Displays, storage devices, power supplies, cooling arrangements, and mechanical protection all contributed to the finished design.
Battery-powered portable computers also developed during the decade. Machines such as the Epson HX-20 and later laptop designs required manufacturers to pay closer attention to electrical consumption, battery capacity, and heat production. Reducing power requirements could improve both operating time and the practicality of smaller enclosures.
By the end of the 1980s, the basic distinction between desktop, transportable, and laptop computers was becoming recognizable, even though the technology would continue changing substantially.
Workstations and Supercomputers: Different Demands
While personal computers were becoming smaller and more affordable, engineering and scientific workstations continued to require more capable hardware. Manufacturers such as Sun Microsystems, Silicon Graphics, and Digital Equipment Corporation developed systems intended for technical computing, graphics, and other demanding applications.
These workstations often used desktop or deskside enclosures with expansion capabilities, substantial power supplies, and active cooling. Their processors, memory, storage devices, and specialized graphics hardware could impose greater electrical and thermal demands than those of ordinary home computers.
The growing use of standardized equipment racks also remained important in professional computing environments. Rack-mounted systems provided a practical way to organize multiple computers, storage devices, and supporting equipment while preserving service access and managing power distribution.
At the highest end of computing, the Cray-2, introduced in 1985, demonstrated a particularly unusual approach to cooling. Its densely packed electronic modules were immersed in Fluorinert, an electrically insulating liquid that carried heat away from the circuitry.
This was not the same as the mineral-oil cooling experiments later associated with computer hobbyists. The Cray-2 used an engineered dielectric cooling system designed specifically for the machine’s performance and packaging requirements.
Immersion cooling carried heat away from components packed too closely for conventional air cooling to be practical. As with the earlier Cray-1, the physical arrangement of the machine was closely connected to its electrical performance.
The workstation and supercomputer markets continued to develop their own solutions to the same power, space, and cooling problems. Different categories of computers continued to solve related engineering problems at very different scales.
By the end of the 1980s, several recognizable computer designs had become established: compact keyboard computers, expandable desktop systems, integrated all-in-one machines, transportables, and early laptops. Internal power supplies and cooling fans were common in expandable PCs, while smaller systems often relied on external adapters or passive cooling. The following decade would bring greater standardization, faster processors, increasingly important thermal management, and a rapidly expanding market for complete and custom-built computers.
The 1990s: Standardization, Expansion, and the Rise of the PC Builder
By the 1990s, personal computers had become established tools for business, education, and home use. The familiar desktop computer continued to evolve, while tower cases became increasingly common as users added larger hard drives, optical drives, expansion cards, and more powerful processors. Computer manufacturers needed enclosures that could accommodate changing hardware without requiring an entirely new design for every model.
The way people purchased and assembled computers changed as well. Large manufacturers sold complete systems, local computer shops built machines from standardized components, and enthusiasts increasingly selected individual parts to create their own configurations. These approaches existed alongside one another, supported by improvements in case design, power supplies, cooling, and hardware standards.
From Desktop Cases to Towers
The horizontal desktop case remained common throughout the 1990s, particularly in offices where a monitor could sit directly on top of the computer. However, tower cases offered advantages as internal components multiplied. By placing the motherboard and expansion cards vertically, manufacturers could provide additional drive bays and expansion space without occupying as much desk surface.
Mini-towers, mid-towers, and full towers offered different combinations of size and capacity. Smaller systems suited ordinary office and home use, while larger towers provided room for additional storage devices, expansion cards, and cooling equipment. Full towers were particularly useful for servers, workstations, and heavily expanded personal computers.
Most PC cases were constructed around a steel chassis with removable metal panels and plastic front bezels. Beige and off-white remained familiar colors, although other finishes appeared. The emphasis was generally on practicality, manufacturing cost, and compatibility rather than displaying the internal hardware.
Drive bays became a defining feature of these enclosures. External 5.25-inch bays accommodated CD-ROM drives, tape drives, and other devices, while 3.5-inch bays held floppy drives and hard disks. Internal drive cages provided additional storage capacity. The number and arrangement of these bays could significantly influence a buyer’s choice of case.
Access to the computer’s interior also became increasingly important. Removable side panels, drive mounting brackets, and standardized expansion-slot openings made repairs and upgrades more convenient. However, many cases still required removing numerous screws, and sharp sheet-metal edges were an unwelcome feature of some inexpensive enclosures.
From AT to ATX: A New Standard for PC Construction
Intel introduced the ATX motherboard and power-supply specification in 1995. Earlier IBM-compatible computers commonly followed variations of the AT and Baby AT designs, which had served the industry for years but presented growing limitations.
AT-style systems typically used two six-pin motherboard power connectors, commonly identified as P8 and P9. These connectors had to be installed in the correct orientation, with the black ground wires positioned together. Incorrect installation could damage the motherboard.
AT power supplies also commonly relied on mechanical switches connected to household AC power. Turning off the computer physically interrupted power to the supply. This arrangement was straightforward, but it limited the ability of the operating system or motherboard to control the computer’s power state.
ATX introduced a different approach. Its motherboard layout reorganized component placement and provided a standardized rear I/O arrangement, reducing the need for separate cables and expansion-slot brackets for common external ports. It also replaced the two AT motherboard power plugs with a keyed connector and provided for software-controlled power management.
With ATX, the front-panel power button typically became a low-voltage control connected to the motherboard rather than a mechanical switch carrying incoming AC power. The power supply provided standby power, allowing the motherboard to respond to the button and, when supported, other wake-up events.
ATX also addressed airflow and component placement, although actual cooling effectiveness still depended on the case, power supply, processor, and installed hardware. Manufacturers and independent builders could now design around a more consistent arrangement of parts.
The transition was gradual. AT and Baby AT systems remained in use for years, and manufacturers continued selling compatible components while ATX adoption increased. By the end of the decade, ATX had become a leading standard for new desktop PCs, helping establish the basic arrangement still recognizable in many modern computers.
Faster Processors and Growing Cooling Requirements
As processors became faster and more complex, cooling became a more visible part of personal-computer construction. Earlier microprocessors could often operate with modest heatsinks or without dedicated fans, but the increasing power consumption of later processors made active cooling increasingly important.
As clock speeds and power dissipation rose, heatsinks became increasingly common on 486-class systems, though cooling requirements varied by processor and configuration. Pentium-era systems made processor heatsinks and fans more routine, turning cooler selection and installation into practical concerns for builders.
A heatsink transfers heat away from a processor through a thermally conductive material, typically aluminum or copper. Its fins increase the surface area available for transferring heat to the surrounding air. A fan improves that process by moving air across the fins and carrying warmed air away.
Thermal compounds were used between processors and heatsinks to improve contact by filling microscopic gaps between their surfaces. Correct mounting pressure and proper contact were essential, since even a capable heatsink could perform poorly if it was not installed correctly.
Processor packaging was changing as well. Some used ceramic or plastic packages mounted in sockets, while others appeared on cartridge-style modules. Intel’s Pentium II, introduced in 1997, used a processor cartridge that required a corresponding retention and cooling arrangement. These changes affected motherboard layouts and the space needed around the processor.
Case ventilation became more important as processors, expansion cards, and storage devices generated additional heat. Power-supply fans continued to contribute to overall airflow, while some systems incorporated additional case fans. Nevertheless, cooling arrangements varied considerably, and not every computer required multiple fans.
On an AT-style motherboard, the two six-pin power connectors must be oriented with their black ground wires together. Unlike a keyed ATX connector, the older arrangement leaves more room for a damaging installation mistake.
Expansion, Storage, and the Problem of Cables
The growing capabilities of personal computers also made their interiors more crowded. Hard drives, floppy drives, CD-ROM drives, sound cards, modems, network adapters, and graphics cards all required space, electrical power, or data connections.
Parallel ATA (IDE) used wide ribbon cables to connect storage devices to the motherboard or controller. Floppy drives used similar ribbon cables, while power connections required separate wiring from the power supply.
The cables did their job, but they made the interior harder to work in. Wide ribbons could obstruct access to components, complicate maintenance, and interfere with airflow depending on their placement. Builders often folded or routed cables around drive cages and expansion cards to keep the interior manageable.
Cable management during this period was generally practical rather than decorative. The immediate goal was to connect the hardware without obstructing fans, preventing panels from closing, or making future repairs unnecessarily difficult. Dedicated cable-routing channels and hidden compartments were not yet common features of ordinary PC cases.
Power supplies also needed to accommodate additional hardware. Standard peripheral power connectors supplied electricity to hard drives and optical drives, while smaller connectors served floppy drives. As systems expanded, the number and arrangement of available connectors became increasingly important.
Case size, power-supply capacity, and room for expansion had to be considered together. Choosing a larger case or more capable power supply was often a practical decision about future upgrades rather than an attempt to improve appearance.
Workstations, Servers, and Enterprise Hardware
Workstations and servers had their own enclosure and power requirements. Their designs emphasized expansion, continuous operation, and service access in ways that ordinary home computers often did not.
Sun Microsystems’ SPARCstation family included compact desktop workstations, while larger deskside systems provided additional expansion and storage capacity. Silicon Graphics offered machines such as the Indigo and Indigo2, whose distinctive enclosures combined industrial design with the practical requirements of graphics-oriented hardware.
Digital Equipment Corporation’s Alpha-based systems similarly appeared in workstation and server configurations designed around their performance and expansion requirements. Their designs also looked quite different from the beige towers common in the PC-compatible market.
In business computing, rack-mounted servers became increasingly important as organizations sought to consolidate equipment and make better use of available space. Standard equipment racks allowed multiple systems to be installed in a relatively compact area while preserving access for maintenance.
Server power supplies and cooling arrangements were often designed for continuous operation. Some systems provided redundant power supplies, allowing a failed unit to be replaced or bypassed without immediately shutting down the computer. Hot-swappable drives and other service-oriented features also appeared in enterprise equipment.
Redundancy and dense packaging brought additional engineering requirements. Densely packed equipment needed sufficient airflow, reliable electrical distribution, and careful planning for heat removal. The physical design of a server installation could be as important to its reliability as the individual computers it contained.
From Local Builders to Mass-Market PCs
The expanding market for PC-compatible hardware supported both large manufacturers and smaller businesses assembling computers from commercially available components. Not all major manufacturers used standard retail parts, however; many retained proprietary case, motherboard, or power-supply designs. Customers could purchase complete systems from major manufacturers, order customized configurations, or visit a local computer shop to have a machine built to their specifications.
Several companies became particularly recognizable during the 1990s:
- Dell expanded its direct-sales and build-to-order business, allowing customers to select configurations without relying entirely on traditional retail distribution. Its model contrasted with traditional retail sales.
- Gateway 2000 became known for its direct-sales model and distinctive cow-patterned packaging. Its approach made direct ordering familiar to home buyers.
- Compaq was a major manufacturer of IBM-compatible personal computers and servers. Its machines ranged from consumer desktops to business systems.
- Packard Bell became familiar to many home-computer buyers through its consumer desktop systems. Its ready-to-use systems were sold widely through retail channels.
- Micron Electronics, later associated with the MicronPC and MPC names, competed in the desktop and business-computer markets. It offered another alternative to the largest PC brands.
- Falcon Northwest, founded in 1992, specialized in high-performance and gaming computers. Its business centered on configurations aimed at demanding users.
- Alienware, founded in 1996, developed a reputation for gaming-oriented systems and distinctive case designs. Its machines helped establish a distinct market for purpose-built gaming PCs.
Other manufacturers, including IBM, Hewlett-Packard, Acer, and NEC, also competed in the expanding personal-computer market. Their systems ranged from standardized desktops to proprietary designs that were less convenient to modify.
Local computer shops formed another important part of the industry. Often described as white-box builders, these businesses assembled computers using motherboards, processors, memory, cases, power supplies, and other components obtained from distributors. A customer could request a particular configuration, have an existing machine upgraded, or bring in a system for repairs.
A small shop could purchase compatible parts from distributors, assemble them, and provide service without manufacturing any of the underlying electronics.
Individual builders benefited from those same standards. Someone comfortable installing expansion cards and drives could increasingly assemble an entire computer from purchased components. The challenge had shifted from constructing electronic circuits to selecting compatible hardware, connecting it correctly, and configuring the finished system.
Proprietary power supplies, cases, connectors, and motherboard layouts could still complicate repairs or upgrades. A PC-compatible label did not guarantee that every component could be replaced with an off-the-shelf part.
Standardized parts made PC assembly more accessible, but compatibility still had to be checked. A matching form factor did not guarantee adequate power, cooling clearance, or freedom from proprietary wiring.
The End of the Decade: Complete Systems and Custom Builds
By the end of the 1990s, many households and businesses purchased complete computers with operating systems and common software already installed. Enthusiasts and independent shops, meanwhile, could choose compatible processors, motherboards, drives, cases, power supplies, and cooling equipment to suit particular needs. Building a PC increasingly meant assembling and configuring existing components rather than constructing their circuits.
These choices were connected. A case determined available drive bays, expansion clearance, and fan positions; a power supply needed enough capacity and the right connectors; and the processor’s cooling arrangement had to fit the motherboard and enclosure. Standardization made those decisions easier, but did not eliminate compatibility problems or proprietary hardware. The growing heat output of faster processors and graphics hardware would make the relationship between enclosure design, electrical power, and cooling still more important in the following decades.
Summary
Computer cases, power supplies, and cooling systems changed as the machines themselves changed. Early computers required substantial infrastructure, with cabinets, electrical distribution, and cooling systems designed around the needs of each installation. As electronic components became smaller, computers could move into laboratories, offices, schools, and eventually homes. Yet every reduction in size brought its own engineering challenges.
Microcomputers changed who could assemble and maintain a computer. Early enthusiasts often assembled circuits, built kits, or supplied their own enclosures and power systems. Complete computers soon became available alongside these approaches, and manufacturers developed increasingly integrated and convenient designs. By the 1980s, personal computers ranged from compact keyboard-based machines to expandable desktops, while professional workstations and supercomputers continued pursuing solutions suited to their different requirements.
By the 1990s, a builder could assemble a useful PC without designing or soldering its circuits. ATX motherboards, compatible cases, standardized power connections, and readily available cooling equipment helped establish a market in which manufacturers, local computer shops, and individual builders could work with many of the same components. As complete systems became commonplace, building a computer increasingly became a matter of selecting, assembling, and upgrading hardware rather than constructing the underlying electronics.
Standardization did not settle every design problem. Faster processors, more powerful graphics hardware, and increasingly compact systems would place new demands on electrical power and heat management. At the same time, computer cases would begin evolving from largely practical enclosures into expressions of personal preference, with new materials, layouts, lighting, and cooling arrangements. Those changes are the subject of the next part of this history.
More from the Foundations of Computation Series
- Foundations of Computation: From Mechanical Systems to Early Electronic Computers
- Interfaces, Storage, and Early System Structure
- Mainframes, Minicomputers and Microcomputers
- Cray Supercomputers
- Computing History and Shared System Design
- 8-Bit Personal Computing and Fragmentation
- IBM PC Platform and Early Standardization
- The Technologies We Carry: Why Some Computers Become Part of Our Story
- BBS to Internet: Local to Global
- Experiencing DOS Today with FreeDOS and VirtualBox
- From Isolated Computers to Local Area Networks
- The Server Emerges as a Role
- When Hardware No Longer Defines the Computer
- UNIX Becomes a Family
- Software Distribution, Freedom, and Licensing
- UNIX: From Source to Standards
- UNIX Beyond the Original Source
- BSD: The First Free UNIX Family
- The Linux Family
- DOS to Windows: Microsoft, IBM, & Novell
- Commodity Hardware Transformation
- Hardware History: CPU & RAM
- Hardware History: Motherboards and Firmware
- Hardware History: Cases, Power Supplies, & Cooling: The Early Years