Slashdot Mirror
The Last of the Punch Card Programmers
Peter Cus writes "Cluny Lace, an English lacemaking manufacturer, has reverted to 19th-Century Leavers machines in order to stay competitive. These 19th-Century machines use Jacquard punch cards. Ian Elm, thought to be the last of the card punchers, says young people don't want factory work: 'Younger people coming into a trade want a guarantee of a career out of it, and this is so uncertain.'"
i work in the metalworking trade, and many times an old screw machine can do a run of parts so much more efficiently than a cnc machine that it is used instead.
Punch cards can be a pretty useful educational tool. In the 1980s I had an intro to computer science class where we had to write our first programming assignment in fortran(*) using punch cards. Second and subsequent assignments would use terminals. The professor explained that doing so was terribly obsolete but that this experience would help us understand why some computer languages (fortran in particular) and some operating systems (including unix) are the way they are. He added that the deck of blank punch cards we would have to buy would also provide us with plenty of book marks for the rest of our years in college.
(*) Fortran was only used in this intro computer science class. This class was required for many engineering and science majors who were more likely to use fortran than computer science majors. Unexpectedly in the mid 1990s I actually used fortran as my company was contracted to move some chemistry software from mainframes to personal computers.
In the 1980s I had an intro to computer science class where we had to write our first programming assignment in fortran(*) using punch cards.
Back in the 1970's when I was in college, the first day of my first computer class the professor told us that "the keypunch machines are down the hall." I asked him, "uh, as in punch cards?" At that point I'd been hacking assembler code on microcomputers for a few years and doing real-world interfacing, and really wasn't interested in punch cards. Sure, had it been a one-time experience like you had, that would have been interesting. But an entire school year spent in front of a keypunch machine, submitting jobs to an IBM 370, when there were rooms full of 3270 terminals all over the place? No thanks. I dropped that class that afternoon.
The higher the technology, the sharper that two-edged sword.
Well, if you mean by "screw machine" a machine that makes screws, well, generally they literally stamp fasteners from spools of metal wire. Much faster and more efficient than trying to machine such parts ... that would be hideously slow in comparison.
Not a stamper. Think of a metal lathe, then porcupine it with multiple cutting tools and power feeds, to get a turret lathe. Then add even more clockwork/gearing and it can make multiple parts pretty much hands off, and you got a screw machine.
http://en.wikipedia.org/wiki/Screw_machine
All that clockwork/gearing is complicated as heck to modify compared to feeding a new gcode file into a CNC. However, being hyper-specialized, if you don't need a slow precisely controlled negative X-axis movement or whatever, a screw machine probably has a big ole high tension spring that moves "instantly" vs the CNC slowly methodically and precisely crawling neg x-axis.
As far as hideously slow, you'd be surprised even in ancient history what a couple horsepower and sharp cutting/forming tools can do...
"Science flies us to the moon. Religion flies us into buildings." - Victor Stenger
Not a stamper. Think of a metal lathe, then porcupine it with multiple cutting tools and power feeds, to get a turret lathe. Then add even more clockwork/gearing and it can make multiple parts pretty much hands off, and you got a screw machine.
Okay. I've seen equipment like that, but the screw machines I'm familiar with (I did a lot of data acquisition work in the fastener industry, many moons ago) were basically large solenoid-operated stamping machines. Rows and rows of the things, all thumping out about three or four parts per second. They had separate dies to form the various parts of the fastener, and were fed by large spools of metal wire (steel, brass, whatever.) They were also very loud, as I remember, although not anywhere near what I experienced in a some automotive stamping plants. Earplugs for the win.
These were mostly self-tapping parts (drill screws and the like) and the systems I developed measured various attributes such as drill time, peak and tapping torque values, end-load, and so forth. This was mostly for statistical process control purposes, although I did a number of laboratory test systems as well. Those were used for design testing, as well as assessing performance of competitors' parts.
The higher the technology, the sharper that two-edged sword.
Modern machines are now being built that are run directly from computers, but I'd say, given that these are huge expensive machines that are often resold and moved to new locations rather than bought new, the majority still run on paper tape.
The issue of quality isn't directly related to the machines being computer-driven. The quality depends on the care of the designer, the 'stitch count' or density, and quality of thread, etc. As with many manufactured goods, you can get lace for less money if you accept lower quality. No surprise there.
I assume the computer-driven machines would let an operator change the stitch count. These days, there are few people (in the West anyway) who know how to create a 'punching' as it is called, and fewer who are interested in learning. Strangely, the remnant of my father's business is just starting to get orders from Asia, so maybe 'Free Trade' is finally coming around to the point where manufacturing costs in the US are competitive with Asia in this regard, but there really is no one ready or willing to meet the manufacturing demand if it ever really comes back. You can probably ditto this sentiment for US shoe manufacturing, furniture, etc.
John Leavers invented those machines in 1812 and they're still in use. If two hundred years isn't job security, I don't know what is!
In those days, UNIX ran on machines that we would consider tiny today, and so it had small input buffers, which you might say it was influenced by the 80-column punched card, or perhaps just by the 32k bytes (or 64k or 128k, if you were rich) PDP-11 system memory size. These buffer size limits were in the kernel, but easier to see in the /bin utilities.
A "screw machine" is the name of an automated lathe, and while they can certainly produce screws, they're often used to make all kinds of cylindrical metal parts. They are not limited to making just screws. We used them to make everything from locomotive fuel injectors to hex-socketed screwdriver shafts.
We also made plenty of screw-threaded items, but never just ordinary "raw" screws. Other, simpler machines, such as the ones you described, had long ago taken over that task.
John