Loading summary
Transcript1 lines
- [00:02]
A
The machinist occupies a special place in industry. Using a set of mechanized tools and drawing on years of experience and vibes, they take something from raw metal to finished form. Machining was part science, part magic. A respected craft that brought pride and a good living to its many practitioners. Then, in the 1950s, a revolutionary new technology sought to replace the machinist's capabilities with a string of numbers. One Japanese company arose to take the fullest advantage of this trend. In today's video, the Rise of Numerical Control CNC and Fanuc. Today's video is brought to you by the Asonometry Patreon. Machine tools are power driven tools that work metal. There are a variety of them. They are generally classified and into two Machining and metal forming tools. Machining tools remove metal from the workpiece in the form of tiny chips. A common example is the lathe. The lathe holds and spins the metal workpiece at a high speed while a cutting tool moves around it, cutting it. Another important one is the milling tool, the lathe's conceptual inverse. Here we have a fast rotating cutting tool and and this time it is the workpiece moving around the cutting tool, shaving off tiny chips. The second major category are the metal forming tools. These units shape metal without removing any of it, most often via deformation, bending, pressing, etc. Such machine tools include forging presses and hammers, hydraulic presses, and so on. Many machine tools first emerged in the United Kingdom and the United states in the mid-1800s for the production of what else? Weapons of war. But the technology and techniques quickly transferred to making sewing machines, bicycles, automobiles, turbines and the like. Measured by output, these machine tools are nothing. In 1986, the whole American machine tool market was valued in at just $2.7 billion less than the American paper bag industry. But these tools play an outsized role in industrialization. Every manufactured item is either made by a machine tool or by a thing made by a machine tool. Let us give them the proper respect they deserve. For many years, these machine tools were manually operated. It was demanding precision work done by skilled machinists in machine shops. Art, art, art, science. The machinist has to know advanced math like algebra and trigonometry. They have to grok metallurgy, how to read technical blueprints. At the same time, all that technical knowledge must be augmented with experience, judgment and just plain feel like an artist. They must envision the part before they can go machine it. Such skills must be learned in a real machine shop over many years under apprenticeship. Machining makes the machinists A key participant in the process. Even if someone else came up with the idea, they must still defer to the machinist experience. Such control and skill was something that the machinists were very proud of. Over time, however, industry has demanded increasingly complex and intricate metal shapes. But with there only being so many skilled machinists, it became difficult to produce identically precise machine parts at scale. Was there another way? Thus the idea of an automatic module controlling the machine tool like as if it were a human operator. In 1946, General Electric and the Gishold Machine Company adapted gunfire control technology developed in World War II to first record a master machinist's motions on a milling machine. Using sensors, the recording is stored onto magnetic tape and then played back with the hope that it can be reused to make many identical parts at scale. It was clever, but ultimately failed. One reason being that the recording was analogous. A continuous waveform, essentially so that recording can't be easily edited. And over time, distortions in the tape's quality caused accuracy to drift. Legacy wise, this particular machine is perhaps most well known today, if at all, for inspiring writer Kurt Vonnegut's debut novel, Player Piano. In 1948, the university MIT got a call from an industrialist named John T. Parsons, founder of the Parsons Corporation. Parsons ran a company making helicopter rotor blades for the Air Force, and this required operators to drill hundreds of holes using a machine tool called a jig borer. To reduce operator error, he figured out how to couple a computer to that jig borer so that you can feed it a series of numeric instructions stored on punch cards, coordinates and tool movements. Essentially, the jig borer then automatically does the jiggy. Then the Air Force proposed to build a plane with a special wing where the skin and ribs are made from a single piece of metal. Parsons questioned the recommendation on how to achieve that. He felt that if he can wire up a milling machine to a computer the same way he did with the jig borer, then that can do it better. The Air force was like Bett and offered a $200,000 contract to build it. So Parsons subcontracts MIT to help execute on this idea of a milling machine controlled by this string of numerical data on paper tape. In 1949, Parsons, MIT and the Air Force signed that contract. The Air Force provided a standard Cincinnati Hydrotel milling machine to modify. MIT researchers William Pease and James McDonough led the project. The development process was difficult. It took MIT a year to come up with a design that Parsons approved of. And during the Build. There was huge conflict. Parsons being the industrialist, he was wanted as few modifications as possible to the Hydrotel milling machine. More modifications raised the budget and and made the tool harder to operate. But MIT was used to working in a military R and D context where budgets are a secondary concern and researchers being researchers, they envision their project as more than just commercializing an idea that is just boring. Rather, they saw themselves as tackling the general problem of crafting shapes out of items. Ergo, they replaced the milling machine's conventional controls entirely with three hydraulic servos, each capable of receiving separate commands and allowing movement in three dimensions. This wayward intellectual exercise apparently and rightly drove Parsons nuts. After six months, the project was just 30% done, but costs had ballooned to over twice the originally projected budget. Financially strained, Parsons withdrew, though he did receive co authorship on the patent. But the Air Force stepped in and with their help, the tool was completed six months later in March 1952. Today we consider the 1952 MIT machine a pioneer. Newspaper articles at the time dubbed it a mechanical brain. In an interview, aforementioned project leaders Peace and McDonough said said that their technology, which they named numerical control, represented a small step towards a new industrial revolution. After the announcement, MIT held educational seminars and demos to promote numerical control technology to American tooling and aerospace companies. Unfortunately, the reception, especially from the tooling companies, was not that great. A history of the Lab by Kyung Soo Paik quoted Brown recalling that the demo made one of the manufacturers so mad to make them write. MIT's president protesting the university wasting resources on such a boondoggle. Some toolmakers from General Motors who saw it brushed it off, saying it's a pretty poor way to build a million Chevrolet fenders. Many years later, Parsons said that MIT's demonstrations were ridiculous because they didn't know machining. In the end, the Air Force did adopt numerical control in a major way. They went on to fund, at considerable expense, MIT and five companies, including Cincinnati Milling and General Electric, to produce NC controller modules integrated into milling machines for their aircraft factories. Early work was difficult. The modules and their machines broke down frequently on the factory floor due to contamination from electricity, dirt, oil and chemicals. Operators had no idea how the thing worked and could not fix it. The controller and machine manufacturers frequently blamed each other. The Air Force did eventually get it to work, and the five NC controller subcontractors then tried selling their modules commercially. But as none of them cooperated with each other, their systems all worked completely differently. The Air Force had also funded a programming language called automatically programmed tool or APT to go along with this NC installation. APT worked and held elegance, but it also greatly contributed to the system's overall costs. Thanks to these efforts, you can say that the United States machine tool industry held a lead in the implementation of NC technology, and there were people, a subset of people, who recognized that NC allowed for certain benefits. However, the long running connection to the Air Force also left people with the impression that NC was inherently unprofitable without extensive government subsidies. A true double edged sword. But did NC have to be so complicated? Can it be done simpler? Across the Pacific, the news from MIT caught the attention of universities and research institutes in Japan. Soon after the 1952 MIT announcement, work on NC began at the Tokyo Institute of Technology, University of Tokyo and the research institute aist. This in turn motivated the electronics giant Fujitsu to tentatively explore the factory automation space. They put together a small research team and selected Dr. Suemon Inaba to lead them. Inaba is described as compact but scholarly looking. I myself might say he's elfish. Born in the Ibaraki prefecture to a long line of village leaders, he apparently spoke Japanese with a noticeably slurring dialect. He studied precision instruments at the University of Tokyo during World War II. After graduating in 1946, he joined Fujitsu as a mechanical engineer, which made him a bit of an odd duck because Fujitsu named Fuji Telecommunications at the time was more known for computers and obviously telecommunications than mechanical engineering. Inaba recalled kicking off in 1955 with very little development started from scratch, beginning with the acquisition of technical papers from mit. The team then got a fortuitous boost from one of Japan's top machine tool companies. In March of The following year, 1956, Tsunezo Makino, CEO and founder of the company Makino Milling Machine Learning, went to India. There he was told that if Japan's machine tools did not have nc, then they were not advanced. Distressed, Makino boldly declared that Japanese machine tools will have NC by the Osaka International Trade Fair in 1958. Upon returning to Japan, Makino scoured the industry landscape and came across Inaba's team, and they quickly struck a partnership. By December 1956, the group effort produced a prototype numerical control device powered by a Sanyo motor and a type of pre transistor electrical circuit called the Parametron. This NC controller module directed a turret punch press imported from the United States. All it did was punch holes into sheet metal, but it was good enough progress to be hailed in the media. Inaba then turns to making an NC milling machine by the 1958 Osaka Trade Fair, they pull it off, though the tool they demonstrated was not commercially viable. After the exhibition, though, Mitsubishi Heavy Industries Nagoya Aircraft Factory is sufficiently impressed with Inaba and Fujitsu to commission an NC milling tool for their F86 fighter jet. Fujitsu, Mitsubishi and Hitachi thus collaborate in in an all star event to deliver the first production NC tool in November 1958. This NC tool was just barely competent. It was so sensitive to electric noise that even a motorbike passing by the factory might cause the device to error out. And each month one of its 100 vacuum tubes blew out, which was extremely inconvenient. In 1959, Inaba and Fujitsu rocketed past domestic machine tool competitors like Toshiba and Mitsubishi with a breakthrough NC enabled milling machine. At its heart is something called an electro hydraulic pulse motor. Inaba invented and patented the motor himself, though he got the core idea from one of Fujitsu's automatic exchange equipment products. How does it work? The name is quite informative. First, a human interprets the blueprints and produces a list of coordinates for the machine to cut. This is put onto a paper tape which is fed into a Fujitsu Fatcom 128 general purpose computer. The Fatcom 128 reads the tape and decides the lines and cut paths. Then another thingy called a director translates that information into a sequence of electrical pulses and transcribes it onto magnetic tape. The pulse sequence goes into a machine control unit. Inside that unit, the sequence controls the flow of highly pressurized hydraulic oil through a special valve. The hydraulic motors thus move a heavy piece of metal along the cutter in a certain direction slowly. The pull sequence allows us to cut the metal into the desired shape. Inaba and his team intentionally designed their tool to be simple but practical. For instance, the 1952 MIT machine ran on a closed feedback loop, meaning that it read and responded to its own real world results, which was technically nifty, but probably contributed to its high cost and impracticality. The Fujitsu machine, on the other hand, was largely open loop, like a fancy player piano that carved metal. Rather than jam out Scott Joplin's the Entertainer. This stripped down version of NC was key in allowing the module to survive factory floor conditions, and it went into the market in 1960. After this first success, one might expect Fujitsu to further pursue the machine tool space. But Inaba instead decided to focus his company on only the NC controller technology, partnering with the Japanese machine tool companies rather than competing against them. Their first open loop NC controller module, the Fanuc 220. Fanuc stands for Fujitsu Automatic Numerical Control, released in 1960, but it struggled in the market. Why? Well, first, it cost about 10 million yen, which was more than the tool itself. Second, it also needed human operators to manually tweak the whole pull sequence to accommodate variance between tools. Makino suggested an offset function that allowed operators to quickly make lasting offsets to fix any tool variance errors. They also suggested removing a curvilinear processing feature and and replacing it with a straight line feature to make it cheaper. This feature deletion as well as transistorization let fujitsu price the Fanuc 260 controller module at 2 million yen, just a fifth that of the Fanuc 220. Sales surged from 60 units in 1965 to 388 in 1968. Back in the United States, one of the key drivers of NC adoption was was the rise of the machining center throughout the 1960s. An early machining center was Kearney and Trekker's Milwaukee Matic Model 2. This $145,000 tool was equipped with a cutter, rotating horizontal table and an automatic tool changer. This tool changer was a rotating drum holding about 30 tools. When needed, an arm can go and equip itself with a new tool from the drum in about nine seconds. Such fast, precise movements can only be performed using numerical control. The NC powered machining center consolidated multiple machining functions into drilling, milling, boring, tapping, reaming and some lathing, though not too well, an entire production center in one. The American machine tool industry for decades led the world and for all those decades. And it thrived on machine tool makers building different tools, knowing that they did different things and that machine shops are going to buy all of them. But NC enabled machining centers leveled and reshaped the playing field, ending niches that once existed in milling, boring, drilling, what have you. So those guys rushed to build their own machining centers, driving NC module adoption and and forcing themselves to compete mano a mano in an increasingly crowded space. But making a good machining center requires new competencies in complicated areas like control theory and electronics. The Japanese toolmakers just bought from Fanuc. But many of the American companies tried to roll their own solution with varying amounts of success. In the mid-1960s, a slowdown in the aerospace industry combined with tightening credit causes the cyclical machine tool industry to crash. Hard sales declined 50 to 60% between the 1967 peak to 1971. This opened the door for financial conglomerates. Private equity essentially to go in and start buying up and merging together these small machine tool businesses. This was detailed in the wonderful book when the Machine stopped, written in 1989 by Max Holland. Backed by Wall street money, these conglomerates paid more than any other machine tool firm. Yet they lacked knowledge or even interest about the machine tool industry itself. Companies like Burgmaster often sold thinking that they can tap pools of cash to fund more products. But it turned out that the conglomerates simply wanted to milk the old cash cow so that they can go buy the next big thing. This starved the US machine tool companies of cash that could have been reinvested into the business, particularly to adopt numerical control. By 1970, Global Machine Tool export share held by American firms declined to just about 15%. The export leader those years was Germany, with about 30% market share, aided in part by Siemens aggressively adopting NC as well as a favorable exchange rate. Fujitsu's NC division did not turn a profit until 1965. They could not have survived the hard early years alone, not only because of the financial support from the Fujitsu giant, but also access to the products of Japan's top computer maker. But by 1971, Fujitsu now controlled 80.7% of the of Japan's NC market, its peak market share. And since NC units earned high margins, the resulting profits were very good. Fujitsu could have kept those profits for itself, funneling cash into increasingly expensive computer R and D or other financial things like what the American conglomerates did. Instead, Fujitsu decided to spin off the NC division into its own company. The parent kept 52% of the shares. Fuji Electric and Germany siemens each got 6%. The latter company got shares due to a cross licensing agreement. The rest of the shares floated on the open market. Inaba, of course, led the newly independent company and would do so for decades. As a businessman, he was a savvy but demanding, daresay, tyrannical leader, known for a love of R and D and alcohol. Though in his later years he apparently gave up drinking for health reasons, he was also a bit quirky. For example, he insisted on being referred to as Doctor of Engineering. And he demanded that all of his workers, including himself, wear yellow and that all his products be colored yellow. Because yellow is the emperor's color or something. Man just likes yellow. You can imagine the NC controller unit as having a brain, which is its computational component and muscles its servo motors. Together they guide the machine tool. In the 1970s, both of these underwent substantial changes. In 1973, Japan was hit by the first of the 1970s oil crises with rising inflation and pricier energy costs and Japanese machine tool customers sought higher precision, greater efficiencies and lower labor costs. NC became a way to reduce setup time, scrap rework and operator labor hours. For instance, it allowed Toyota and Honda to manufacture 20 to 30% fewer parts than their American car counterparts. But as this happened, it started to become clear to Fanuc that that its original open loop electro hydraulic pulse motors were not meeting customer needs. For one thing, it used a lot of oil and power, two things that got very expensive after 1973. And hydraulic systems were proving to be too awkward for precision work. By comparison, direct current electrical motors were rapidly improving thanks to the introduction of a game changing solid state electrical valve called the thyristor. Yet Fanuc owned this pulse motor space. They basically invented it. And the patents protecting it were a key competitive advantage. Inaba even once told a friend that to have it taken away from him would be like taking away his own life. But it had to be done. To manage the transition, he took a dual track, ordering development on an electro hydraulic pulse motor that did not use oil as well as partnerships for direct current servomotor technology. When the former proved to be extremely noisy, he went to the latter with no sentiment. In mid-1974 Fanuc struck a licensing deal with a small American firm in Wisconsin called Geddes Manufacturing for their closed loop electrical DC motors. The first electric motor based NC tools hit the market in September 1974 in and were a revelation. But it was the controller modules brains where Fanuc made the greatest impact. NC systems used hardwired logic systems that were not all that flexibly changed. But the emergence of the minicomputer in the early 1970s made possible NC systems that were soft wired meaning a programmer can first create and edit the machine instructions inside a digital computer, then feed them directly into the machinery without the use of paper, tape or any of that. They called this computer numerical control or CNC. In 1972 Fanuc introduced the Fanuc 200A, their first CNC device. Three years later in 1975, Fanuc integrated one of Intel's early 8bit microprocessors to produce the Fanuc 2000C. This model suffered some reliability problems early on, but it showed the way. This finally led to the company's game changing 1979 product, the System 6. Powered by the 16 bit Intel 8086 microprocessor plus a cavalcade of custom LSI chips, the System 6 reduced discrete components by 30% which made it cheaper to produce and sell. Fanuc enhanced the line further by making it modular, making it easy to integrate the System 6 into a wide variety of machine tools, from lathes to machine centers and beyond to get CNC capability. In doing so, turning it into the industry standard low cost Fanuc CNC modules armed a rabble of Japanese machine tool makers, enabling them to flood the world market. Thanks to them, Fanuc entered the 1980s as Japan's and perhaps the world's leading maker of CNC controllers, with over 50% market share. As early as the 1960s, people argued that NC would deskill the workers in machine shops, turning them into low paid peons, while management further centralizes its grip over the means of production. Research on this notion over the following decades, and the transition did indeed take decades, paints a nuanced picture. There are entire theses on it, so I'm not going to try to get into it all. But generally speaking, the individual machinist's job did not deskill as originally feared, but it did indeed change. Operators no longer exert intricate and complicated physical effort to craft something out of metal, but after the NC revolution, operators now oversee expensive, highly complicated systems that break down in complicated ways. Fixing and keeping these systems online becomes a new, valuable skill, one more intellectual and conceptual than its priors. And considering the surge in productivity it leads to, it pays probably most likely better. That being said, we do lose that sense of self and pride from literally making something with our hands. I think that does mean something. Alright, 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.