This is great. They replace the software entirely, provide hardware schematics and timing diagram for interfacing, figure out the font format and devise their own fonts. Finally they add a driver to troff - I always wondered at the Mergenthaler Linotron 202 being mentioned in the troff manual.
Note the cost of the Linotron 202 is implied to be relatively inexpensive at $45k without options.
Ken Thompson wrote a disassembler on day one so they could examine the machine code.
Reading about all the hardware problems they had getting a printer to work was rather amusing. As I sit here waiting for the photocopier repair man to turn up it's illuminating to contrast the giant steps which have taken place over the last 40 odd years regarding the admittedly extremely difficult technical problem of printing black text and images onto pieces of paper.
I remember the 202, the sound of the photo paper feeding, and the cracking sound the floppies made at certain times--was it when one was saving fonts to them?
I worked as a systems/software engineer at Linotype-Paul in Cheltenham UK, which is where the 202 was designed and built. I worked on tbe hyphenation programs on ut, and wrote a large part of the application used on the Linoscreen 7000 book pagination terminal that was used to drive paginated books into a 202. Tbe Linoscreen 7000 was a derivative of the Datek 7000, aquired (including myself) from Datek systems in Wembly North London. Mergenthaller was our US AGent. I used to have my own 202 taking up a full third of my office. They where incredably reliable, so this story here is an unusual tale. A definate lemon machine.
As you say, they were very reliable. I had a lot to do with 202s over several years, and don't remember any problems that resulted from the 202 itself.
"Since we began using this program instead of Mergenthaler’s
formatter, most of our “hardware” problems have disappeared.
...
There is also a memory diagnostic sufficiently better than
Mergenthaler’s that the repairman asked for a private copy (which we
provided, on paper tape)."
Reading this part made me think, "of course! It was written by some of the most brilliant minds at Bell Labs." Many compaines would be lucky to have had software for them written by Bell Labs.
Considering what a piece of junk the 202 and the software running it turned out to be, Mergenthaler would have done better to just give the 202 to Bell Labs for free and opened up all their hardware and software specs to them, on the condition that the Bell Labs team rewrote their software, and made suggestions for improving the hardware, or even paid Bell Labs for the privilege. That would have been a win-win for both companies.
That story has been repeated many times with Linux/BSD device drivers; expert kernel developers are much better at writing drivers than anyone working for hardware companies.
Linotype/Mergenthaler shipped 1000's of 202s, they where considered to be very reliable. I used one for about 8 years and never had any problems at all with mine. It sounds like these guys got a lemon machine.
So did phototypesetters basically work like printers, or was the output just used as a "master" to duplicate more copies of things like books using another kind of machine?
There's references to the 202 requiring "silver nitrate paper"; basically a (presumably page-sized) photographic negative. That would have been used to produce a printing plate (in the Gutenberg sense) and then used to make actual hard-copies.
The 202 ran out "galleys", photo paper maybe eight inches wide. The typical workflow was a) typesetter at a keyboard enters the file, b) program maybe tidies the file up for hyphenation & justification, c) galley run out and proofread, d) corrections entered if needed, e) galleys cut up and pasted onto cardboard pages, f) photographic negatives made from cardboard pages, g) lithographic plates made from the negatives, h) the pages printed.
Items c and d could repeat a couple of times as necessary, and the customer usually had a look at the product as well ("bluelines" produced from the negatives).
Computerphile interviewed one of the writers of this paper in a series of videos, who explained the brilliant and exiting history of this paper and its recreation, as well as the history of Phototypesetters.
About ten years earlier, in 1970, at Chi Corporation (a spinoff of CWRU) in Cleveland, I used the prototype Harris Intertype phototypesetter. We were one floor below the Harris Intertype Printing Equipment Research Center, and they'd built the first CRT-based phototypesetter. It had a custom CRT one line high and about 12 inches wide, which wrote letters onto photographic paper. The photographic developer, with rollers and tanks of chemicals was built into the machine, so it emitted developed pages. The machine was its own darkroom, with a door so that people could walk inside. This was the prototype of what became the Harris 7000.
Input was via a magnetic tape drive, and fonts were sequences of strokes.
John Langner at Chi wrote a program called Procrustes, which took in fonts in some grid format and turned them into a sequence of vectors. There was no need for reverse engineering; this was done with full cooperation from Harris Intertype. There was also a primitive markup language and layout program, all text-driven. All this ran on an UNIVAC 1108 mainframe. Tapes were written on the mainframe and carried to the phototypesetter. The end result was quite good. I still have some manuals printed from its masters.
We also had the first electrostatic printer, built by Clevite/Brush Instruments.[1] This was a small machine which ran paper over a row of tiny pins which could be turned on and off. This placed an electrostatic charge on the slick coated paper, which was then pulled over a tank of liquid containing toner. The toner was attracted to the charge, and was then somehow fused to the paper. The result was more like dark grey on light grey than black on white. Resolution was under 100DPI. This was a prototype.
Al Misek at Chi built a controller for the thing so that it could talk to the 1108. The controller had an early character generator, so it could print text in one rather bad font. It could also do graphics, I wrote software to do 3D plots with it, and still have some of the printouts. I think that's one of my drawings in the Computerworld article [1]; my 3D graphing program was often used as a demo.
Clevite/Brush didn't intend to build a printer. They built chart recorders (like this [2]) to draw plots on paper from analog input data. (That's how you recorded data from experiments and industrial processes back then.) Those devices moved a pen back and forth mechanically, so they had very low frequency response. The electrostatic printer technology was developed for a faster chart recorder. Then someone realized it could do more than draw lines, so they made up a prototype and let Chi use it.
Clevite never did much with that technology, but Versatec did, and for most of the 1970s, electrostatic printers were a popular output device. Quality never got very good, but the machines were fast and not too expensive. "Troff" supported Versatecs.
20 comments
[ 3.3 ms ] story [ 58.2 ms ] threadNote the cost of the Linotron 202 is implied to be relatively inexpensive at $45k without options.
Ken Thompson wrote a disassembler on day one so they could examine the machine code.
Read it! It's full of goodies.
Considering what a piece of junk the 202 and the software running it turned out to be, Mergenthaler would have done better to just give the 202 to Bell Labs for free and opened up all their hardware and software specs to them, on the condition that the Bell Labs team rewrote their software, and made suggestions for improving the hardware, or even paid Bell Labs for the privilege. That would have been a win-win for both companies.
Items c and d could repeat a couple of times as necessary, and the customer usually had a look at the product as well ("bluelines" produced from the negatives).
1. https://youtu.be/CVxeuwlvf8ww 2. https://youtu.be/XvwNKpDUkiE 3. https://youtu.be/HdModNEK_1UU
http://www.eprg.org/papers/202paper.pdf
Input was via a magnetic tape drive, and fonts were sequences of strokes. John Langner at Chi wrote a program called Procrustes, which took in fonts in some grid format and turned them into a sequence of vectors. There was no need for reverse engineering; this was done with full cooperation from Harris Intertype. There was also a primitive markup language and layout program, all text-driven. All this ran on an UNIVAC 1108 mainframe. Tapes were written on the mainframe and carried to the phototypesetter. The end result was quite good. I still have some manuals printed from its masters.
We also had the first electrostatic printer, built by Clevite/Brush Instruments.[1] This was a small machine which ran paper over a row of tiny pins which could be turned on and off. This placed an electrostatic charge on the slick coated paper, which was then pulled over a tank of liquid containing toner. The toner was attracted to the charge, and was then somehow fused to the paper. The result was more like dark grey on light grey than black on white. Resolution was under 100DPI. This was a prototype.
Al Misek at Chi built a controller for the thing so that it could talk to the 1108. The controller had an early character generator, so it could print text in one rather bad font. It could also do graphics, I wrote software to do 3D plots with it, and still have some of the printouts. I think that's one of my drawings in the Computerworld article [1]; my 3D graphing program was often used as a demo.
Clevite/Brush didn't intend to build a printer. They built chart recorders (like this [2]) to draw plots on paper from analog input data. (That's how you recorded data from experiments and industrial processes back then.) Those devices moved a pen back and forth mechanically, so they had very low frequency response. The electrostatic printer technology was developed for a faster chart recorder. Then someone realized it could do more than draw lines, so they made up a prototype and let Chi use it.
Clevite never did much with that technology, but Versatec did, and for most of the 1970s, electrostatic printers were a popular output device. Quality never got very good, but the machines were fast and not too expensive. "Troff" supported Versatecs.
[1] https://news.google.com/newspapers?nid=849&dat=19700225&id=Q...
[2] http://www.ebay.com/itm/Clevite-Brush-Instruments-15-6327-10...