Electronics Guide

Tracing the Lineage of Electronic Innovation

Most electronic devices are descendants. The transistor in a phone traces back through the planar process to the point-contact device Bell Laboratories demonstrated in December 1947, and that device in turn descends from the crystal detectors radio amateurs built decades earlier. A technology genealogy follows those lines of descent: it asks what a given device inherited, from which parent, and what changed at each step. The result is neither a list of inventions nor a simple timeline. It is a map of dependencies.

The genealogical metaphor is useful because electronics evolves in ways that resemble descent with modification. A single device branches into many uses, as Lee de Forest's three-electrode Audion of 1906 branched into amplifiers, oscillators, and switches, and from there into transcontinental telephony, broadcast radio, radar, and the first electronic computers. Separate lines converge, as computing, radio, imaging, and inertial sensing converged in the smartphone. Branches also terminate: magnetic core memory, magnetic bubble memory, and the plasma display panel each had a working generation and no successors. Reading a technology as the product of its ancestry explains constraints that otherwise appear arbitrary, from the persistence of the 8-bit byte to the survival of the RS-232 signaling conventions in industrial equipment.

This category examines six lineages in depth, each chosen because it runs the full length of the field and still shapes practice today. Beyond those six, the sections that follow describe the shapes that genealogies take, the enabling technologies that gate progress along them, and the ways engineers and students can use genealogical reasoning as a working tool rather than a historical curiosity.

Articles in This Category

The articles below trace six long lineages from their origins to their present state, following the branch points, replacements, and convergences along each path.

How Lineages Branch

Branching is the most common pattern in electronics genealogy: one device turns out to serve several unrelated purposes, and each purpose develops its own line of specialized descendants. The vacuum tube illustrates the pattern clearly. The Fleming valve of 1904 rectified; de Forest's triode of 1906 amplified. Amplification alone produced the repeaters that carried the first transcontinental telephone call between New York and San Francisco in January 1915, the audio chains of broadcast radio, and the sound-on-film systems of the late 1920s. The same tube used as a switch produced digital computing: ENIAC, completed in 1945, contained 17,468 tubes.

Other branches of the tube family diverged so far that the kinship is easy to miss. Karl Ferdinand Braun's cathode-ray oscilloscope tube of 1897 became both the television picture tube and the laboratory instrument display. The cavity magnetron that John Randall and Harry Boot developed at the University of Birmingham in 1940 turned the tube into a high-power microwave source for radar. Traveling-wave tubes and klystrons became satellite and radar transmitters and remain in production, because no solid-state device yet matches them for output power at millimeter wavelengths. The transistor replaced the tube in small-signal work and left these high-power, high-frequency branches largely intact.

Branch points usually occur where a device acquires a new property rather than simply a better number. Gain made the triode a different technology from the diode. Switching speed, not amplification, made the transistor decisive for computing. Recognizing which property drove a branch helps explain why an apparently inferior technology sometimes wins a market: it satisfies a property the incumbent cannot supply at any price.

Convergence and Cross-Pollination

Lineages also merge. The smartphone is the standard example, combining a microprocessor, a cellular radio, a solid-state image sensor, a capacitive touch panel, microelectromechanical inertial sensors, a lithium-ion cell, and a satellite navigation receiver in one enclosure. None of those components was developed for a telephone. Convergence became possible only when each contributing line independently reached acceptable size, power, and cost, which is why the integrated result appeared decades after its parts.

Cross-pollination between fields is equally common, and often accidental. Percy Spencer, working with an active magnetron at Raytheon, observed that it melted confectionery in his pocket; Raytheon filed a patent on microwave cooking in 1945 and put the first commercial oven, the Radarange, on the market in the late 1940s. The charge-coupled device that Willard Boyle and George Smith conceived at Bell Laboratories in 1969 was intended as a shift-register memory, and it found its lasting application in imaging instead, from astronomy to consumer cameras. Satellite navigation, built for weapons delivery and force positioning, became a civilian utility after the United States discontinued the deliberate degradation known as Selective Availability in May 2000.

Procurement can act as a cross-pollinating force in its own right. In 1963 the MIT Instrumentation Laboratory's purchases for the Apollo Guidance Computer absorbed roughly 60 percent of United States integrated circuit production, and by 1965 the Minuteman II missile program had overtaken Apollo as the largest single customer. Neither program invented the integrated circuit, but their willingness to pay early prices funded the yield learning that made commercial parts affordable. The military-industrial electronics complex and technology transfer and espionage categories examine that mechanism in more detail.

Enabling Technologies and the Long Wait

Genealogies stall when a concept arrives before the technology that would make it practical. The gap is often measured in decades. Friedrich Reinitzer described the liquid crystalline state in 1888; RCA demonstrated a working dynamic-scattering liquid crystal display in 1968; thin-film-transistor active-matrix panels displaced the cathode-ray tube in computer displays only during the 1990s, once large-area transistor fabrication on glass became economic. Ching Tang and Steven Van Slyke published the thin-film organic light-emitting diode at Kodak in 1987, and OLED panels reached volume production in phones roughly two decades later, after encapsulation and material lifetime problems were solved.

The enabling advance is frequently a process rather than a device. Jean Hoerni's planar process at Fairchild, and the monolithic integrated circuits that Jack Kilby demonstrated at Texas Instruments in 1958 and Robert Noyce designed at Fairchild in 1959, mattered less as inventions than as manufacturing methods that could be repeated with rising yield. Later steps followed the same rule: chemically amplified photoresists, chemical mechanical planarization, copper interconnect, and extreme ultraviolet lithography each unlocked a generation of designs that were already understood in principle. When tracing a lineage, the question "what became manufacturable?" usually explains timing better than "what was invented?"

The practical consequence is that a stalled idea is not necessarily a bad one. Artificial neural networks, proposed in the late 1950s and largely dormant after the limitations of single-layer perceptrons were publicized in 1969, returned to prominence once training algorithms and inexpensive parallel hardware arrived. Reviewing dormant branches against current capability is one of the more productive uses of genealogical thinking.

Abandoned Branches

Some lines end. Magnetic core memory dominated computer main storage from the mid-1950s until semiconductor memory, beginning with the Intel 1103 DRAM in 1970, undercut it on cost and density; core survives only in the vocabulary, in phrases such as "core dump." Magnetic bubble memory attracted heavy investment in the 1970s as a non-volatile successor to disk, then lost decisively to falling disk and semiconductor prices. Plasma display panels, invented at the University of Illinois in the 1960s, held the large-format television market for a period and were withdrawn by their last major manufacturers in the mid-2010s, beaten by liquid crystal panels on cost and power.

Extinction in a genealogy rarely means the technology stopped working. It means a competitor improved along a dimension that mattered more to buyers, usually cost per unit of function. Studying terminated branches is therefore less about failure than about the criteria that actually decide adoption, a theme developed further under failed technologies and obsolescence.

S-Curves and the Limits of Extrapolation

Technologies commonly improve slowly at first, then rapidly, then slowly again as physical or economic limits bind. Plotting a performance measure over time produces the familiar S-curve, and successive S-curves stacked end to end describe a lineage. The value of the model is predictive in only a limited sense: it warns that the steep segment will end, without saying when.

Semiconductor scaling supplies the clearest case. Gordon Moore observed in a 1965 article in Electronics that the number of components per integrated circuit was doubling roughly annually, and revised the rate to about every two years in 1975. The observation held for decades because it described an economic target that the industry organized itself to meet. The related scaling rules that Robert Dennard and colleagues set out in 1974, under which power density stayed constant as devices shrank, stopped holding in the mid-2000s. Clock frequencies plateaued in the low gigahertz range, and the industry turned to multiple cores, specialized accelerators, and advanced packaging to keep delivering performance. The lineage continued; the mechanism driving it changed.

Two lessons follow for anyone reading a trend line. First, an exponential in a technical measure usually rests on an economic arrangement, and it lasts only as long as that arrangement does. Second, the successor to a saturating technology is often already in the laboratory during the plateau, which is why the search for it should begin before the incumbent visibly fails.

Other Lineages Worth Tracing

The same method applies well beyond the six articles in this category. Display technology runs from Braun's cathode-ray tube of 1897 through the liquid crystal and plasma panels to organic light-emitting and microLED devices, with each transition driven by a different constraint: depth, then power, then color and contrast.

Wireless communication descends from spark-gap transmitters, which radiated across a broad band and could not be tuned sharply, through continuous-wave alternators and vacuum-tube oscillators to modern digitally modulated cellular systems, the fifth generation of which was first standardized in 3GPP Release 15. Energy storage runs from Volta's pile of 1800 through Gaston Planté's lead-acid cell of 1859 and nickel-based chemistries to the lithium-ion cell Sony commercialized in 1991, whose energy density set the boundary conditions for portable electronics ever since. Each of these threads appears throughout the history of electronics categories, viewed there by period rather than by lineage.

Using Genealogical Thinking

For practicing engineers, ancestry explains constraints. Knowing why a design choice was made, what alternatives were rejected, and what problem an earlier generation was solving prevents the reinvention of discarded solutions and identifies inherited assumptions that no longer apply. Legacy interfaces, register maps, and protocol quirks are usually intelligible only in terms of the hardware they were written for. A genealogical reading also indicates where an incumbent technology sits on its S-curve, which is a practical input to architecture and sourcing decisions.

For students and general readers, genealogies convert electronics history from a list of names and dates into a connected argument. Following the logic from crystal detector to point-contact transistor to planar integrated circuit makes each step comprehensible as a response to a specific limitation rather than an isolated stroke of insight. That perspective also sets realistic expectations about the future: the technologies of the next decade will, in the main, be modified descendants of the ones already in laboratories today, and the future perspectives and emerging trends category examines the most credible of those candidates.