Selectric Typewriter conversions

From: Arlen Michaels <arlen_at_acm.org>
Date: Tue Feb 4 15:33:01 2003

on 4/2/03 3:36 AM, Eric Smith at eric_at_brouhaha.com wrote:

> IBM made "I/O Selectrics," which had heavy-duty mechanisms intended
> for continuous output. Model numbers that come to mind are the
> 1052 and 753, though my memory regarding these seems more than a
> little fuzzy.

For what it's worth, my memory (no less fuzzy than Eric's) remembers that as
a model 735.

But even with the more robust "I/O Selectrics" there was a hidden gotcha:
when you sent data to the mechanism you had to time it right. The Selectric
needed varying amounts of time to tilt and swivel the type element,
depending on the letter chosen. To cope with that, lots of the early
computer interface projects just stuck a software delay loop in the driver
code to wait plenty of time for each letter to finish printing before it
sent the next character code to the Selectric. This apparently caused a lot
of mechanical wear-and-tear as the clutch relentlessly cycled in and out,
and even the heavy duty "I/O" mechanism would eventually crap out from all
the lurching. The right way to do it was to make sure your driver polled
the state of some microswitches that IBM had thoughtfully buried in the
clockwork to provide feedback on the progress of the print cycle. That way
you could start sending a new character at a magic moment just _before_ the
previous print cycle fully finished, which provided the fastest possible
printing and extended the life of the clutch.

Like John said, VooDoo.

Regards,
Arlen Michaels
Received on Tue Feb 04 2003 - 15:33:01 GMT