99 Years Ago This Month
Published in ESD January, 2006
For novel ideas about building embedded systems (both hardware and firmware), join the 35,000 engineers who subscribe to The Embedded Muse, a free biweekly newsletter. The Muse has no hype and no vendor PR. Click here to subscribe.
By Jack Ganssle
Scientific American devotes a page every issue to a retrospective on articles it ran 50, 100, and an astonishing 150 years ago. IEEE Computer appropriately recaps pieces 16 and 32 years old.
Embedded Systems Design doesn't follow this policy. We haven't busted the century mark, yet, but will in only 83 more years. Then I'm sure developers busily cranking out a billion lines using some abstraction we can't image will be amused to read of 64 bit processors, programming languages, transistors and non-telepathic communications protocols. That assumes, of course, that reading hasn't been replaced by 3-D imaging via brain implants.
So permit me to look back to an article we didn't run, since we didn't exist, none of us were born, and the world-changing implications of the invention described by that non-article weren't understood at all.
99 years ago this month Lee De Forest was granted patent 841,387 for the vacuum tube (' valve' to our English colleagues). The tube could arguably be called the most important invention in the history of electronics, for it introduced for the first time the concept of an active element. Tubes could amplify, rectify and work as logic components. Though transistors eventually supplanted these devices for most applications, there's really nothing that a transistor can do that a tube can't.
Before tubes there were no consumer electronic products. No cell phones, stereos or even monos; wax cylinders (and to a lesser extent in the early years) disks transmitted music via a needle coupled directly to a loudspeaker-like horn. This technology was entirely mechanical and was devoid of a single wire.
There was no radio, at least for consumers. Commercial operators manipulated a telegraph key that induced very high voltages into big coils which discharged in a sizzling spark, creating broadband electromagnetic energy transferred to an antenna. It wasn't till 1901 that Marconi was able to harness this crude technology to transmit the letter ' S' across the Atlantic. He received (or possibly imagined he'd heard the three short dits) the broadcast on Signal Hill in Newfoundland using a kite-born antenna and several million watts of transmitter power. I remember standing there, overlooking a sea strewn with icebergs, amazed at Marconi's audacity and success. Since the transatlantic cable had been carrying telegraph traffic for 35 years to nearly the same landing in Newfoundland one wonders why Marconi bothered to build such an elaborate bit of gear that performed so much worse than the cable. Yet when tubes came along his vision changed the world.
Before tubes telephones had limited reach. Without an amplifier signals degraded quickly. Until Michael Pupin's 1900 invention of the loading coil, which passively boosted transmissions, calls were limited to a couple of hundred miles. His coil increased the range to around a thousand miles, far short of that needed to link this gigantic country into one telephonic net. Though telegraphy transmitted messages across the continent, the personal contact of a voice call that binds families, communities and nations had to await Lee De Forest's invention.
Edison, Swan, and a host of others worked feverishly in the waning years of the 19th century to perfect an electric light bulb. By 1880, after trying some 6000 compounds as filaments, Edison still searched for a material that wouldn't burn out in a handful of hours. Working with carbonized filaments of every sort of plant, he noticed that the bulb's glass envelope developed a shadow near the positively charged side of the filament, suggesting that carbon was somehow propelled through the vacuum from the negative terminal to the plus side. He found that the carbon was charged, suggesting a flowing current.
But Edison was a busy man, by 1882 creating the entire infrastructure needed for electric lighting, from generators to transmission lines to electric meters. He put aside his observations for a time, but in 1883 had built a version of the bulb with a third element, a platinum plate, to measure the effect. Indeed, current flowed from the filament to the plate across the vacuum. None of these experiments reduced the buildup of carbon on the bulb, so, being a practical man he moved on to other projects.
The effect was a mystery. In 1884 at Edison's urging professor E.J. Houston wrote a paper that was published on the very first page of the very first issue of the Transactions of the American Institute of Electrical Engineers, released at the very first meeting of the AIEE - a predecessor of today's IEEE - which described but didn't explain the effect. He wrote that he couldn't conceive of electricity flowing across the vacuum to the platinum element. This was long before Einstein's 1905 paper on the photoelectric effect, which, to his eternal chagrin, ushered in the age of Quantum Mechanics.
Edison had for all intents and purposes invented the diode, but didn't understand what he was seeing. The flow of current is now called thermionic emission but was long known as ' the Edison Effect.'
I'll present my Better Firmware Faster seminar in Melbourne and Perth, Australia February 20 and 26th. All are invited. More info here. The early registration discount ends January 20.
While working as a consultant to Marconi he pondered the thorny problem of, as he put it, detecting the ' feeble to and fro motions of electricity from an aerial wire,' and realized his diode was the perfect solution. Tests confirmed his thesis and by 1905 he had US patents on the device.
Meanwhile other researchers were looking into the nature of the atom. Joseph (later Sir Joseph) Thomson discovered that that the flow of electricity inside a vacuum was a stream of negatively-charged particles smaller than an atom; particles he named ' corpuscles,' though we now call them ' electrons.' Thomson measured a constant charge to mass ratio and deduced that these corpuscles all had a small but fixed charge.
Enter Lee De Forest, engineer and inventor, who had been tinkering with various forms of telegraphy. For reasons that I can't discover, he modified Fleming's tube by adding a small coil of wire between the filament and the platinum plate.
Cue choir of angels. De Forest's modified tube, which he called the Audion, was the first triode, a three element electronic component that could amplify tiny signals. The patent was issued in January of 1907, 99 years ago this month.
De Forest didn't understand how the Audion worked. But he did go on to use one in the first electronic amplifier in 1911, and then invented a multi-stage amplifier using a number of Audions the following year.
The inventor went on to a checkered and brilliant career, suffering through many business failures and more lawsuits. He broadcast the first radio program with commercials, and even received an Academy Award in 1960 for his inventions that gave motion pictures sound.
So what were the implications of the vacuum tube?
For one, it revolutionized telephone communications. With amplifiers now possible, true long-distance phone calls became, if not common, at least attainable. I remember even in the early 60's how an interstate phone call was an Event which took precedence over everything. The first transcontinental phone line opened in 1915, just 4 years after De Forest built his amplifier. Many quickly followed.
Computers changed from slow and crude relay logic to true electronic digital machines that operated at, for the time, blinding speeds. At Bletchley Park British engineers built the 1500 tube Colossus in 1943 to break WWII German Enigma codes. The 1946 Eniac used 18,000 tubes, and could perform 385 multiplications per second. That's about what the 8008, the first 8 bit microprocessor, could do in the 70s.
But perhaps nothing changed so quickly as radio. In 1914 Edwin Armstrong patented the regenerative receiver, which used a single vacuum tube to massively amplify microvolt-level RF signals. Patent battles then weren't much different than those today; the courts eventually ruled that Lee De Forest's similar 1916 patent had priority. Tubes were very expensive, costing $5 to 8 or more (a lot of money back then) so this single-tube circuit won a lot of followers.
But regenerative circuits are unstable. In 1918 Armstrong won a patent for the superheterodyne radio, which changed the world. Even today, most radios use this circuit.
Called a superhet for short, these radios mix the incoming RF with that of an oscillator whose frequency is a handful of megahertz away from the signal. That mixing process produces both sum and difference frequencies. A filter rejects the sum; the difference, probably 10.7 MHz for your FM radio currently playing in the background, is called the ' intermediate frequency,' or IF. It's easier to amplify the 10.7 MHz instead of the signal's natural 100 MHz or so. Double conversion superhets (the most common kind) repeat the process, creating another typically 455 KHz IF.
No radio manufacturer took the superhet seriously till amateur radio operators managed two way transatlantic communications using this circuit. In 1924 RCA introduced the first consumer superhet, and managed to sell 148,000 in the very first year.
As has so often happened in the last century, this new design which needed lots of expensive tubes created a market for, well, tubes. Manufacturers found ways to drastically reduce their costs so that by 1930 regenerative receivers were obsolete. Six tube superhets became the norm as tube prices plummeted.
But the courts stepped in again, and ruled that Frenchman's Lucien Levy's superhet patent had priority over Armstrong's.
Over the years engineers perfected many variations on the original triode. Tetrodes added a second grid to minimize plate to the grid capacitance which induces oscillations. Pentodes added yet a third grid to control electrons that bounced off the plate.
World War II spurred much more research into these marvels of electronics. Both sides of the battle pursued the elusive invention of RADAR, but it proved difficult to generate a reliable source of high frequency microwaves. In great secrecy the British perfected the cavity magnetron, a vacuum tube that generates vast amounts of microwave energy. The magnetron insured Allied RADAR was far superior to that of the Axis.
Small receiving tubes were common in consumer and industrial appliances into the 60s. Old-timers remember Heathkit which sold hundreds of electronic projects in kit form: TVs, ' hi-fis,' ham radio gear and more. We geeks found each new catalog from the company more interesting than Playboy (almost). Their OM-2 oscilloscope used only 13 tubes ' it was comprised of, to use modern terms, essentially 12 transistors and one CRT. Can you imagine designing anything today with less than a million transistors?
Look around your house. You'll still find some vacuum tubes soldiering on. The microwave oven has a magnetron. High end guitar amplifiers eschew transistors for tubes; purists note a cleaner, purer tone. And of course the CRT, whether in your TV or computer monitor, is nothing but an overgrown vacuum tube using electrons sprayed from a heated cathode to paint pictures on phosphor.
One could rightly claim that the electronics revolution started 99 years ago this month. Though a century is a long time, it took millennia to go from the wheel to the tin lizzy, only 40 years to progress from tubes to transistors, and then just over a decade to get to the integrated circuit.
My 3 pound laptop has billions of transistors embedded in a handful of ICs. The equivalent tube circuit ' which couldn't be built ' would fill a city. Yet each of those little transistors does no more than a simple triode.
What an amazing world we engineers, and our predecessors, have built!
It takes a lot of energy to accelerate electrons across the tube, so typically several hundred volts are applied to the plate. And the cathode won't spew electrons till it's hot, creating the ' warm up' period familiar to anyone using a CRT monitor or TV. Ah, I remember the soft glow of a dozen filaments in the back of any electronic appliance in the olden days. And the heat! An old radio warmed the room; ancient tube computers required monstrous air conditioning equipment.
Filaments were resistive heaters usually driven by 6.3 or 12.6 VAC, though the range ran from a volt or two up to 117. This bit of history lives on in the datasheet for Signetic's Write Only Memory chip, a 1972 April Fool's joke. One spec lists Vff at 6.3VAC... for the filaments, what else?
A further note:
Triodes are considerably easier to understand than transistors. A cathode is heated by the filament, though on some tubes these two elements are combined into one. The heat boils electrons off a coating on the cathode. The plate is always positively charged by a power supply, so the electrons speed in that direction. But to get there they pass through the grid, a lose wire coil.
Consider the effect of applying a small negative charge to the grid. Electrons heading towards it will be repelled, to lesser or greater degree depending on the size of the charge. So a small tweak in grid voltage can control a lot of electron flow. The tube amplifies weak grid voltages.
Figure 1: A triode. To the right, a 12AX7, a very popular miniature (!) tube containing two triodes, sharing one filament. Picture on the right courtesy of Audiomatica (http://www.mclink.it/com/audiomatica/tubes).