The Electron: A Biography, Part 2: From the Vacuum Tube to the Data Center

From early radio and television assembly to surface mount technology and AI data centers, the evolution of electronics manufacturing reveals how innovation, miniaturization and automation have transformed the industry.
by Ronald C. Lasky, Ph.D.

Last time (with a lot of help from my student researcher Rishaan Nalwar), I left off with a bit of an irony: engineers had been soldering telegraph wires, running telephone lines and lighting up entire cities with electrons for the better part of eighty years before physicists like J. J. Thomson and Robert Millikan actually proved what an electron was. Science had finally caught up with engineering. Now it was time to put this newly confirmed little particle to work in a whole new way.

As electric lights and appliances became more common, radio was emerging. Initially, hobbyists used lead sulfide (galena) crystals to catch weak radio signals drifting through the air. The crystal needed no amplification, but the signal was so faint that you needed headphones pressed right to your ear to hear anything at all. Lee de Forest’s triode vacuum tube, shown in Figure 1, changed everything: those whisper-quiet signals could now be amplified enough to drive a speaker, and the whole family could gather around and listen together. Surprisingly, radio did not take off overnight. It took nearly a decade for anyone to realize that advertising, not subscriptions, would pay for it all.

Radio

By the 1930s, the radio had become an indispensable part of American life. During the Great Depression, radios served as a primary source of news, entertainment, and connection for many families. Despite financial struggles, the affordability and value of radio made it a staple in American households: by 1930, over 40% of American households owned a radio, and by the end of the decade, that number had more than doubled to 83%.

The radio’s role extended well beyond entertainment. It was a tool for unity and reassurance. President Franklin D. Roosevelt’s Fireside Chats gave comfort and a direct line of communication to citizens during the hardest years of the Depression and, later, World War II.

Manufacturing all of these radios gave birth to the electronic assembly industry as we know it. A radio like the 1924 Atwater Kent set shown in Figure 2 would have been assembled entirely by hand, wire by wire, with solder and flux. Figure 3 shows a modern reproduction of that kind of early radio circuit board up close, and Figure 4 shows the same kind of bar solder and flux that would have been used to build it, not so different from what is used today. Radio assembly began in earnest in the late 1920s, and under the demands of World War II it had become a genuinely disciplined, methodical process, formalized by new military specifications (MIL-SPECs) that gave the whole industry a rigor it had not had before.

Television

Through the 1950s, radio gave way to television, and electronics assembly grew right along with it. The widespread adoption of the printed wiring board in the 1940s and 50s finally did away with the rat’s nest of point-to-point wiring inside a radio or TV chassis. Hand soldering hung around for a while, but by the 1950s, wave soldering had taken over, using the same bar solder and fluxes, we use today, shown in Figure 4. Wave soldering would go on to become the backbone of electronics assembly for the computer industry that followed.

The shift from radio to television did more than change how boards were built, though; it reshaped American culture. Radio had been an imaginative, auditory medium: families gathered around and listened, picturing the events and characters for themselves. Television added moving pictures, and suddenly national events, entertainment, and advertising all became far more vivid and immediate. It nudged American culture toward a shared visual experience, standardizing tastes and values across the country the way only a nightly sitcom or the 1960 Kennedy-Nixon debate could. Advertising became more powerful too, tying products to images of modernity and success, even as television quietly pulled people away from local theaters, radio dramas, and other community-based entertainment, and further into home-centered, suburban life.

Television Transforms Electronics Manufacturing

Television did something else, too: it forced electronics manufacturing to grow up. Radios were relatively simple and could be hand-wired point-to-point by a skilled technician without too much trouble. Televisions were a different animal entirely, with dozens of vacuum tubes, transformers, coils, and capacitors, all packed together at high voltages and radio frequencies. Point-to-point wiring became slow, error-prone, and impossible to scale, so manufacturers turned to printed circuit boards, which offered repeatable layouts, faster assembly, and much easier troubleshooting.

TV design also encouraged a modular approach: a set might be broken into an RF tuner, an IF strip, an audio amplifier, and separate deflection circuits, each built on its own board or sub-chassis. This modularity made assembly and repair simpler and let multiple teams build and test different sections in parallel. Labor itself became specialized. Rather than one technician building an entire set, individual workers handled narrowly defined tasks: inserting components, soldering, or inspecting. It was a real assembly line, and it made electronics manufacturing accessible to a much broader workforce than the radio days had required.

Because televisions ran at higher voltages and frequencies than radios, marginal solder joints could cause failures you could literally see and hear: a flickering picture or a crackling speaker. So manufacturers had to get serious about soldering consistency, flux quality, temperature control, and inspection, which were really the first stirrings of the process control and quality assurance we take for granted in electronics manufacturing today. Jigs, fixtures, and in-line testing became standard, holding boards in place during assembly and checking voltages and signal response at every stage, rather than waiting until the very end to catch a problem.

Television production was still mostly hand-assembled, but its scale and complexity paved the way for what came next. When transistors, and later integrated circuits, replaced the vacuum tube, the modular, PCB-based habits the television industry had already built made it easy to bring in automated component insertion and wave soldering. In a real sense, television manufacturing was the bridge between handcrafted electronics and the automated board assembly we know today, and it directly enabled the mass production of early computers, military electronics, and eventually the PC and the smartphone.

Miniaturization: Not Just the Smartphone’s Story

It is easy to assume the smartphone is responsible for the relentless miniaturization of electronics. It certainly demands more functionality crammed into less space than anything before it. But that is not quite where the story starts.

Miniaturization actually began in the military, aerospace, and medical worlds of the 1950s and 60s, where weight, reliability, and power consumption were life-or-death concerns. Missiles, spacecraft, hearing aids, and portable radios all pushed the development of transistors and integrated circuits, and proved what was technically possible. But these were low-volume, specialized products. They did not yet create the sustained, high-volume commercial pressure needed to make dense electronics a mainstream expectation.

That pressure showed up more clearly in the consumer camera industry of the 1970s, 80s, and 90s as shown in Figure 5. Compact 35mm cameras were already packed tight with optics, shutters, film transport, and batteries, yet buyers increasingly expected autofocus, automatic exposure, motorized film advance, and electronic flash, all squeezed into the same small body. Meeting those expectations required dense, low-power electronics tightly integrated with mechanical and optical parts, and camera makers, especially in Japan, became some of the earliest and most aggressive adopters of custom chips, flexible circuits, hybrid assemblies, and miniature motors and sensors. Cameras were genuinely mass-market, global products, and in that environment, miniaturization stopped being a curiosity and became a competitive necessity.

The shift to digital cameras in the 1990s only raised the stakes: image sensors, high-speed signal processing, memory, and displays all had to fit into a handheld device running on a small battery. A lot of the design tricks that later defined the smartphone, like system-on-chip integration, aggressive power management, and dense packaging, were worked out first inside a camera body.

So rather than being the origin of electronic miniaturization, the smartphone is really its culmination. It inherited a design culture built over decades, first by the military and medical worlds, and then, just as importantly, by the humble camera.

Why Surface Mount Technology Took Over

Surface mount technology, or SMT, did not take over circuit board assembly because of one clever invention. It won because through-hole assembly simply could not keep up with everything the industry was demanding of it at once: more components, better electrical performance, and efficient mass production.

Miniaturization was one big driver. Early radios and TVs, built with point-to-point wiring or through-hole parts, worked fine when circuits were simple, but as integrated circuits packed in more capability, through-hole components, which need drilled holes and generous lead spacing, started eating up board space and limiting components to one side of the board. Surface mount parts could sit directly on the board’s surface, at much higher density, and on both sides at once, which mattered enormously for calculators, personal computers, and eventually anything meant to fit in a pocket.

Electrical performance mattered too. As clock speeds climbed, the long leads on through-hole parts introduced parasitic inductance and capacitance that muddied signals and increased interference. Surface mount components, with their short or leadless connections, largely solved that problem and became a natural fit for the performance demands of modern chips.

Then there was plain economics. Through-hole assembly depended on manual labor for component insertion and soldering, which was hard to scale and hard to keep consistent. Rising labor costs and rising volumes pushed manufacturers toward automation, and SMT enabled fully automated lines: pick-and-place machines, solder paste printing, and reflow soldering, all faster, more repeatable, and far more amenable to real process control than anything that came before.

None of this would have worked without supporting technology catching up too. PCB fabrication improved to support finer traces and more reliable vias. Solder pastes and flux chemistries matured for precise deposition and consistent wetting during reflow. Vision-guided placement equipment and automated inspection cut defects further. And as IC pin counts climbed, packages like SOICs, QFPs, and eventually ball grid arrays let manufacturers pack in hundreds or thousands of connections in a compact footprint, something through-hole packaging simply could not do.

Finally, once high-volume consumer electronics created the economic case for automated SMT lines, and military and aerospace applications validated it for weight and reliability, the outcome was not really in doubt. Through-hole technology still has its place for certain mechanical or high-power jobs, but SMT became, and remains, the foundation of modern electronics manufacturing.

Today, and Some Thoughts on Tomorrow

It is worth stepping back and appreciating just how far we have come. For most of human history, right up through Andrew Jackson’s time, communication and travel changed hardly at all. If you had lived in Boston and been invited to Jackson’s 1829 inauguration, the invitation would have taken two weeks to reach you, and a good, fast stagecoach trip to Washington would have taken about another week. Twenty-five years later, thanks to the telegraph and the railroad, a message could arrive almost instantly, with the traveler following a day behind.

Our own era’s version of that leap is arguably even more dramatic. The first electronic computer, ENIAC, shown in Figure 6, managed about 500 floating-point operations per second, cost the equivalent of about $8 million in today’s dollars, ate as much electricity as 100 homes, and broke down every day or two whenever one of its 18,000 vacuum tubes gave out. Today’s smartphone has more than a billion times ENIAC’s computing power and a billion times its memory, all for under $1,000.

And we are not slowing down. A single rack of Nvidia’s newest DGX GB200 supercomputer draws 120 kilowatts, enough to power around 100 homes, and delivers 1.44 exaflops of performance, more than 100,000 times what is in your smartphone, for about $15 an hour to run. Data centers built to house racks like these are now being planned at the 10-gigawatt scale; at ten cents a kilowatt-hour, that is on the order of a million dollars an hour, just to keep the lights, or rather the electrons, flowing.

It is a fun bit of trivia that, over an average American lifetime, roughly ten to the twenty-fifth power electrons pass through a given cross-section of wire to support that lifestyle. That is about how many electrons each of us “uses” in a lifetime, a genuinely staggering number for a genuinely tiny particle.

Conclusion

The electron may be the single most important thing nature has given us. It provides every chemical bond that holds matter together, from the simplest virus to the largest star, and, as this pair of posts has hopefully shown, humanity has spent the last two centuries putting it to increasingly remarkable use: the telegraph, the telephone, the light bulb, the radio, the television, the computer, and now the AI data center.

We are standing at the edge of some genuinely new territory: artificial intelligence, quantum computing, autonomous vehicles, and technologies we probably have not even named yet. AI is beyond impressive on a screen, but it has not yet shown it can do much in the physical world; autonomous vehicles still seem years away from really arriving; and quantum computing, at least for now, still feels like a solution in search of a problem. Time will tell how all of this shakes out. But if the last 200 years are any guide, electricity, and the humble electron carrying it, will remain the indispensable infrastructure underneath whatever comes next.End of article content

Ronald C. Lasky, Ph.D., P.E., is a senior technologist at Indium Corporation and a professor of engineering at Dartmouth College.