Question Ancient computer peripheral: the acoustic coupler

[80251]

I only saw one of these devices when I was in middle school in the "computer lab". It was used to connect to a remote host via a dumb, green-screen terminal. I don't even know how fast the connection was, but I'm sure it was painfully slow - maybe 300 baud. Practically anyone could type faster than the connection could respond. Much more sophisticated modems that came out some 15 years later could xmit @ 56.6kbs -- and those modems could cost upwards of $200 -- ~$463 in today's dollars. Acoustic coupler: https://www.computinghistory.org.uk/det/41824/Anderson-Jacobson-A211-Acoustic-Coupler/

 

[Paperdoc]

Yeah, I used one of these in the 1970's. I'm pretty sure 300 baud is right.

Most people do not realize that the famous IBM Selectric typewriter with the interchangeable "golf ball" type sphere was designed as a computer I/O device that would print much faster than a normal typewriter, but was smaller and quieter than a Line Printer. I used terminals based on that design for accessing a system in the APL language running on a central IBM 360 /50 machine. That language was designed for interactive use, not punched card submission. Normally several terminals were in one central room for general use. But we had one by itself in a Chemistry Dept. undergraduate lab, using one of these modems and a landline phone to access the mainframe.

 

[80251]

I learned to touch type on the IBM Selectric. I was one of only two guys in my high school typing class. Those typewriters were really expensive back in the day.
An IBM 360, that's going back a ways. Did you ever see one of the infamous IBM 30/30 "Winchester" hard drives?
I remember the industrial sized dot matrix printer the mainframers had, that printer seemed to always have print jobs running.

 

[Paperdoc]

Expensive, yes! Selectrics performed very well and were popular in business and in academia especially because you could change fonts by changing the "golf ball" type head. They also could be a bit costly for supplies because most used a single-use ribbon cartridge that gave beautifully sharp type but only used once through the cart. Multi-use fabric ribbons also were an option.

Actually, the "Winchester" design of hard drives was new in the '60's, but is still the basic design concept today. The unique feature was the idea that clearance of the head above the spinning hard drive disk surface was NOT achieved by supporting the head rigidly over the surface with precise tolerance. Instead, it depended on the phenomenon that the rapidly-spinning disk surface created a flow of air following the surface, and this flow could support the head if the head were held over on a flexible arm. The resulting clearance could be VERY small and self-adjusting for small variations in the disk surface and did NOT require very precise arm construction. The small clearance also made for strong magnetic coupling, allowing faster data transfer. The only restriction (easily met) was that this all worked only when the disk is rotating, so starting and stopping could be done only when the head arm was retracted to a "parking track" location where a "head crash" (contact with the surface) was tolerable.

I do not remember whether this hard drive design was in use in that mainframe system in the late 60's. I do know that an earlier form of random-access storage was in use. There were a few large floor-standing machines. Into each one could place removable "disk packs". This unit was a set of many rigid double-sided magnetic disks about 12"diameter, similar in size to an LP phonograph record, all mounted on a vertical shaft perhaps 10" high. The entire pack was stored in a case to keep it free of dust, and removed from that to place into the drive unit. The drive had arms for each set of heads which positioned them rigidly at fixed clearances over each disk surface. The disk pack rotated, and the arms moved to track locations on the disks just as current HD's do it. These units did NOT use the "air cushion" technology of Winchester drives, but did have the feature that the interchangeable disk packs could allow each drive to serve very different data sets. For example, in the particular case I cited above, the interactive APL system could not be run under the normal OS of the mainframe. So to get it up for users, the disk pack for APL operation was mounted in a drive and the entire mainframe switched over to running that OS and language for a scheduled time period. Then the process was reversed back to "normal" multi-use operations under a different OS. I am not sure what the storage capacity of one disk pack was. In the system I used, each user was allotted several workplaces of 32 KB each, and there surely were a few hundred users' worth of data stored. In any one workspace a user would store their software and their data as named variables, and could copy programs or data variables from one workspace to another.

For those who never encountered it, APL was an acronym for "A Programming Language" (how original!) and was very mathematics-oriented. All data were basically vector arrays (n-dimensional!), so simplest data were a single constant (0-dimensional), a vector of several values (1-dimensional), a flat array (2-dimensional). Larger arrays merely had more dimensions and therefore indices. There were lots of symbols (mostly Greek letters) used for pre-defined operators you used to write your own custom programs or just simple do-it-now calculations. Almost all of these were designed to operate on scalar or vector variables, but the built-in functions in general used matrix algebra techniques suited to n-dimensional arrays. Named pre-written programs could be called as subroutines just by specifying their name and the numeric arguments they were to operate on. It had lots of power and features many non-math people did not use or understand.

Oh, by the way, the common Line Printer used in computer centres at that time was NOT dot matrix. It printed a lot faster than those, and used a different technology. If you want a description, post back here. This is NOT related, of course, to your original post.

 

[80251]

So the industrial printer the mainframers used wasn't dot matrix? That makes sense because that printer burned through paper a lot faster than any single-head dot matrix printer could've worked. I'd B interested in reading more about the Line Printer. I remember it made quite a racket too -- even w/the cover in place.

I once talked to a legal word processor -- she laughed at my 40WPM typing skills and stated if you weren't typing a hundred WPM in her profession you weren't working. She made a $100/hr. and this was back in the late 1980's so that's more like $277/hr. today! I'm almost certain she was using some type of computerized word processor rather than a typewriter by the late 1980's.

 

[Paperdoc]

My understanding of the mainframe systems Line Printer was this. It worked with tractor-feed paper (the stuff with holes punched in tear-off strips along the edges) for fast paper movement. The paper was typically about 15" wide. It printed virtually a whole line in one pass. It had a continuous strip of cast metal blocks, each with one character, that could be pressed against the paper with an ink-impregnated ribbon to make a letter image. Same principle as an older typewriter, but the letters were not on individual arms. The strip of letters MAY have been some linked units, or perhaps a set of sliding blocks in a channel. The entire loop contained three complete sets of alphabetic letters (upper case only) plus one set of special characters, and it rotated continuously across the width of the paper and back around. Behind that strip a head scanned across the sheet and back carrying a hammer. (I don't think there was a sheet-wide array of multiple hammers, but I may be wrong.) Timed pulses to the hammer would knock the right character block forward to print an image in the right space of the paper. Horizontal letter spacing was fixed at 10 characters per inch. The rapid hammer blows and the paper feed system made a lot of noise, so the printer had a sound-absorbing cabinet to help. It really was amazingly fast and tended to be in constant use. The OS inserted pages with labels at the beginning of each job. The system operators would remove the stack of output from time to time, split it up into user packets, and put it out for pickup with the original desk of punched cards submitted for the job.