Loading summary
Transcript1 lines
- [00:03]
A
In December 1957, BusinessWeek profiled a new device from General Electric. The author begins by A new device in a pea sized package stands a good chance of hitting the electrical industry as thunderously as its older and more famous relative, the transistor, struck electronics. How often does a business article hail some new technology like this? And and how often does the thing actually live up to such words? The article was talking about the silicon thyristor and it did indeed revolutionize the electric industry. In this video, the discovery and nearly 70 year impact of the first solid state power electronics device. This video is brought to you by the Astronometry Patreon A generator in a power plant produces an alternating current because a generator is at its heart a wire loop rotating at a constant speed inside a magnetic field. Simply speaking, this pushes the electrons in that wire to create a flow of electricity. In the first half turn, a side of the loop comes up through the magnetic field, pushing the electrons in one direction. So then in the next half turn, things flip in the other direction, creating the undulating wave of alternating current. But our devices cannot use AC power as it is. Transistors expect power to be supplied at fixed levels in a single current direction. AC's reversals mess with that, the transistors will break down, the capacitors bust. And so much other technology requires power, but precisely controlled. For instance, LEDs take in variable levels of power to produce more or less light. And power fed to electric motors or mechanical systems need to be adjusted up or down so it can do useful work. So before power can be used, it must first be converted, fine tuned and controlled. That is what power electronics do. In 1901, Peter Cooper Hewitt invented the first power electronic the glass bulb mercury arc rectifier. A rectifier is what we call a device that turns AC to dc. Its evil sibling is the inverter making DC into ac. Hewitt was the grandson of Peter Cooper, who founded the college Cooper Union. The younger Cooper Hewitt was also an inventor. His most famous invention was probably the mercury arc lamp, a predecessor of the fluorescent lamp. He also had an impressive mustache. Before Cooper Hewitt's rectifier, the only way to convert AC to DC was through electromechanical means. The most well known such device was the rotary converter or just rotary. This basically was an AC driven motor mechanically coupled to a DC generator. They were huge, noisy, expensive and not that efficient by modern standards. And just the very thought is a bit weird. We use electricity to turn a motor so it can turn a generator, eh? After Cooper Hewitt discover that his mercury vapor lamp allowed a current to flow in only one direction. And he built the mercury arc rectifier as an alternative. It has two electrodes, a cathode and anode together inside a sealed envelope of mercury vapor. The cathode here is a pool of liquid mercury. It emits electrons. The anode is some carbon or metal coated with graphite. It can receive electrons but cannot emit them. During the positive half cycle of the alternating current, we get a negative voltage at the cathode and positive voltage at the anode. This causes the cathode to emit electrons that drift over to the anode forming an arc and current. Then when the alternating current flips, we get a positive voltage at the cathode and a negative voltage at the anode. Except the anode cannot emit electrons, so no arc and current forms. The flow is blocked. In essence, the rectifier is a one way street to turn an AC current into a DC one. That street basically chops off the negative half cycle like the leaves off a fat daikin. These devices found immediate use in industrial processes like aluminium reduction and and electroplating. They were also incorporated for lighting incandescent lamps. GE used them to build the first DC power distribution lines in 1905. These mercury arc rectifiers evolved into tube devices called dyrotrons. These descend from argon filled vacuum tubes. They have three an anode, a heated cathode and and sitting in between the two, a control grid. All sealed inside a glass tube full of a gas like mercury vapor, hydrogen or xenon. This control grid springs from the original vacuum tube or triode and is the thyratron's key difference from the mercury arc rectifier. The rectifier was a passive device. So rectification happened automatically per the AC cycle. The thyrotron, on the other hand, is an active device. The control grid blocks the electrons from traveling to the anode during both the negative and positive half cycles of the trip. When we activate the grid with a positive voltage, it creates an electric field that ionizes the gas. The ionized gas lets electrons cascade from the cathode to the anode. The with little loss of voltage. Once the gas ionizes, the tube stays on the grid alone. Cannot turn it off anymore. In other words, it latches open. This lasts until either the main power supply is removed or the voltage at the cathode side goes to zero. Which is what happens when the AC flips the beat current direction. This latching behavior is a big Deal, because it means the user doesn't need to expend further energy to hold the gate open. Thyrotrons first emerged in the 1920s, and what made them so special was that their control gate let you precisely control when to open up and let a current flow through the tube. And this lets you set up timed big bursts. For instance, During World War II, Thyrotrons gained their stars and radars. They helped provide the big bursts of power that allow magnetrons to send radio wave pulses for radars to detect stuff. And they're still used for some of that today. So thyrotrons show their worth. But being glass tubes full of hot gas at first, mercury, later hydrogen, they were fragile. They switched slowly and were also power hungry. Was there something better out there? Then came the transistor, the first solid state device made from modern semiconducting materials. Famously, Bardeen and Brattain at Bell Labs were looking for something else when they came across the first transistor. They were originally looking for a solid state semiconductor device that worked via the field effectively. To be more clear, they wanted a solid state device that creates an electric field. If you gave its control grid a small voltage, that field then motivates electrons to rush across. The grid itself should not draw much of a current. So basically a vacuum tube all in behavior, but make its solid state. It didn't work because the surfaces of semiconductors held traps, and that blocked electric fields from penetrating into the bulk of the material. A true field effect transistor would not come for some more years. So what the Bell Labs duo actually discovered in 1947, the point contact transistor ended up being quite different. It was made from two gold contacts. The emitter and the collector pressed down very close together onto a block of germanium. This block of germanium is attached to the third electrode, which we would now call the base. The germanium is also specially treated. The bulk of the germanium is classified as N type germanium, meaning that it has been doped to have a plurality of electrons, that there is a very thin surface layer that has been treated to be P type, meaning that it has a plurality of electron holes. During operation, the transistor takes in an input current at low impotence at the emitter and uses that to control a larger output current at the high impotence collector. This control produces gain. So the first transistor was an amplifier, but not one controlled by a voltage like a vacuum tube. Rather, it was controlled by another current, that one current we are injecting into the collector. The point Contact transistor is nothing like a bipolar junction transistor, which more looks like a sandwich. But looking back, we can map its operation to that of a PNP bipolar, despite it not having bipolar like junctions. And I say that because of what happens next. As Bardeen and Brattain continued to work on the point contact transistor in 1948, they came across a quirky phenomena. Early point contact transistors did not produce the type of current that can be fed into external devices. So engineers wanted a way to vary the collector's resistance values. The Bell Labs people started tinkering with the collector using heating pulses from a capacitor. Trying to get those values, they discovered that this caused the output current to be unexpectedly high. They called this structure the Hooke collector. And nobody could explain why this over amplification was happening until William shockley theorized in 1950 that the heating pulses formed what is called an inversion layer at the point of the collector. An inversion layer is basically what you get when an electric field inverts semiconductor materials. Type. So if the emitter area was originally P type with lots of electron holes, forming an inversion layer there means repelling enough holes and attracting enough electrons to form a local zone of N type material on top. The inversion layer thus injects additional charge carriers into the collector that were not originally passed in from the emitter, thus causing the off target amplification. Creating this inversion layer at the final P of the PNP point contact transistor meant a different transistor structure entirely. The pnpn. Shockley turned out to be right. The Hooke collector transistor was indeed a PNPN device, and the man was brilliant to have recognized it. However, it would be the Bell Labs research scientist Joule Jim Ebbers who produced the first theoretical model of the PNPN switch. Ebers tragically passed away in his 30s from what was probably cancer. But in 1952, he published a paper which conceptualized the switch as a larger circuit consisting of two stacked and interconnected transistors. An NPN bipolar junction transistor on top of a PNP transistor at rest, both transistors are turned off. Their cumulative junctions block currents from passing through, save a small bit of leakage. But his models showed that under certain conditions, the two transistors will egg each other on like teenage boys at a cliff's edge. Such a positive feedback loop leads to them both jumping off, which means the circuit turns on and stays on. Which is to say that it latches the same latching behavior as we saw in the thyrotron, the current passes through the circuit with ease until explicit effort is taken to close it. In 1954, John Maul at Bell Labs recruited a team to try and make a silicon PNPN switch by following Eber's theoretical model, which today has both their names, the general idea being to use it in some Bell telephone switching device. Note the use of silicon versus germanium. Germanium works fine for research, but its small band gap and low melting point made it impractical for most applications. But producing silicon transistors proved to be challenging. One member of the Ebbers team was the legendary Nick Holoniak, who is perhaps most well known for inventing the semiconductor laser diode, a predecessor of the led. Holoniak argued for hacking the thing together with an existing transistor product. The Mall insisted on making the silicon PN PN transistor switch from scratch, which required then new silicon tech like diffuse junctions and oxide masking. They do it, inventing a whole array of new silicon processing tools and techniques along the way. After a few silicon prototypes, they publish the results. In 1956, news starts getting around the industry that Bell Labs had made something kind of special. In 1954, William Shockley leaves Bell Labs behind and decamps to warm Palo Alto. To build Shockley semiconductor lab, he hires a cadre of very smart, educated men. Twelve of his original 20 employees were PhDs in science or mathematics. But what were they going to make? One thing the lab worked on was figuring out how to produce silicon transistors. The first type were double diffused silicon transistors. The term diffused tells you a bit about how these are made. You melt down pure silicon, carefully diffuse dopants into the melt, and then use that to make transistors. I'm trying so hard right now not to turn this into another 45 nanometer node video. Shockley's other project was a PNPN diode, which he called a four layer diode. This is a simple sandwich of four alternating materials with just two physical connection points or terminals, the cathode and anode at each end. Like the thyrotron, the diode can switch from a high to low resistance mode. But unlike the thyrotron, it does this only when the voltage breaches a trigger breakdown voltage. Imagine something like water coming down a dry stream bed until it hits a small dam. It stays there being blocked by it until abruptly pouring over the top. The device's simplicity apparently fascinated Bill Shockley, and he pursued it at all costs. Yet the device was also fiendishly difficult to produce. Shockley not only wanted to make the silicon transistor itself, but also the raw silicon prior to it and the packaging after it. It presented a massive test base for things to go wrong. By early 1957, the Shockley lab employees were generally turned off by Bill's management style, which seemed to have greatly worsened after he won the Nobel in 1956. Thus, in 1957, eight of his best employees decide to break off and start Fairchild Semiconductor. The traitorous eight. For all his human failings, Shockley was a superlative teacher. Having taught them the core of semiconductor physics, he seeded the beginnings of what we now call Silicon Valley. As for Shockley Semiconductor Lab, the lab eventually completed and launched their Shockley four layer transistor diode in April 1958. The commercial impetus for the project was that AT&T might use the diode for its telephone switching operations. But the Shockley diode was too difficult to use to gain a wide audience in the late 1950s. Silicon production techniques were too inconsistent and this led to varying voltage thresholds or leakage current induced thermal issues that so energized the transistors that they stayed on despite all efforts to turn them off. The Shockley diode, despite its elegance, was not it. Customers needed a way to control the switch. Who was going to build the thing to do that for them? General Electric had provided electrical equipment for decades, but in the mid-1950s they started expanding their work in semiconductors. Among the team's leaders was Bill Gutswiller, who GE hired right out of Marquette to investigate and rate various germanium based solid state rectifiers. Gutswiller became an early convert to the incredible potential of solid state devices in the power industry, eventually earning the name Mr. G.E. rectifier. He pushed the engineers and physicists at GE to produce a home run device. This device would be a solid state rectifier that converted AC power to DC power. Moreover, it would grant the user control over when the rectification happens. Just like the grid controlled Dyratrons. And if it would be made from silicon, germanium rectifiers were already in the market and already replacing some thyrotron tubes. But if you wanted the rectifier to perform more reliably, then it had to be silicon. Put these together and we have the name of the thing. Gutswiller Silicon Controlled Rectifier. He called it the holy grail and urged the company to build it. Fortunately, bell Labs showed GE the way. The 1956 mall paper caught the attention of Gutswiller and his team and they immediately recognized that the PNPN structure detailed in the paper had applications in the power electronics space. Gordon hall and other members in the GE semiconductor lab started playing with various semiconductor techniques. By now, single crystal silicon techniques had improved. So by following the recipe listed in the 1956 Bell Labs paper, GE team produced a prototype four layer alternating PNPN silicon sandwich. However, with this one, a gate was added in the second P between the two N type layers, giving it three terminals in total. Sending a small voltage into the gate will trigger breakdown and let a current through, no matter the voltage level. This basically gave the SCR the same functionality as the Thyratron, perhaps less elegant than the Shockley diode. But it worked and that was what customers actually wanted. Gutswiller recalled in a 2005 oral history. Hall bringing him the first SCR prototype and saying now show me what you can do with it. So so then he the monkey was on my back. I thought of what might best demonstrate the device's capabilities and decided on a motor control. During lunch hour, I walked down to Western Auto, the only hardware store in the village of Clyde, and bought an electric hand drill. In those days, hand drills had two speeds, on and off. With this drill in hand, I went back into the lab and concocted a simple firing circuit for the SCR's control. Leading connected Gordon's new laboratory device up with the motor and the control circuit and turned the power on cautiously. Nothing exploded as I turned the control potentiometer. The drill started turning slowly and then faster as I turned the control further. It worked. The world's first solid state motor control. Up until then, this had all been done on company downtime without any official sanction. But Gutswiller got management on board to accelerate development and the SCR was announced in October 1957, whereupon it gains immediate attention. Besieged for requests of more information, Gutswiller begins writing articles about the product's potential. A month later, Nikolianak left Bell Labs to join General Electric, working to turn the SCR from a toy to into a robust control device capable of switching powerful currents. His work eventually produced the Triac, a device capable of conducting both positive and negative currents upon demand. It was a massive success and other variants quickly entered the market afterwards. Westinghouse soon followed GE in 1959 with their solid state thyristor, which they called the Trinister. L.F. stringer and L.R. tresino in 1966 coined the name Diristor, which comes from Diratron combined with transistor. That name eventually stuck and all the devices in the family, including the scr, got called that solid state devices like the silicon thyristor truly revolutionized power systems the same way that solid state transistors revolutionized logic computing. It founded the modern power electronics industry. Replacing fragile glass tubes with rugged silicon helped make devices far more reliable. Moreover, as Gutswiller's experience hinted, the fast switching thyristor let us slice power at specific spots to get variability that was previously unachievable. With glass, we can now vary motor speeds or brighten or lower light brightness. The solid state power electronics legacy continues with high bandgap materials like gallium nitride and silicon carbide. Our phone chargers and EVs are full of beefy transistors switching power currents on and off. Just like the transistors and CPUs switch data signals on and off. Despite all those new guys on the block, dyristors are still used today, some at gigawatt scale applications like high voltage DC. Imagine that. 60 plus years and still running. Long live the thyristor revolution. Alright everyone, that's it for tonight. Thanks for watching. Subscribe to the channel, sign up for the Patreon and I'll see you guys next time.