28 May
1999
28 May
'99
12:45 p.m.
On May 28, 10:37, Arlen Michaels wrote:
Subject: RE: What's a "computer console"
selectric called?
> From: Don Maslin [SMTP:donm@cts.com]
> On Thu, 27 May 1999, Arlen Michaels wrote:
> > Actually, the biggest challenge in
interfacing this thing to a
computer
> was
> > to sort out how to read one particular status signal from one of the
> > microswitch contacts in the print mechanism, so your computer could
> start
> > sending the next character at just the right moment before the
> mechanical
> > cycle completely finished. Else your software had to pause a few
>
> Didn't you also have to feed it EBCDIC instead of ASCII in order for it
to
> 'understand' what you wanted printed?
I don't think the Model 735 Selectric could even
handle EBCDIC directly.
I
seem to recall the electrical interface was defined as
tilt-and-rotate
signal names. My Selectric terminal certainly didn't do any translation
by
itself from character-codes to solenoid signals, at
least not from ASCII.
I
had to do translation myself before sending to the
printer. One way
would
have been with hardware between the computer and the
Selectric, eg- using
an
eprom to translate each ASCII code into the correct
combination of
Selectric
tilt-and-rotate signals. My lazier way was to simply
put a look-up table
in
my driver code, to intercept each ASCII character
enroute to the printer
and
translate it into the appropriate pattern of solenoid
signals first.
Imagine if you had to drive a dot-matrix print head with raw pin-driver
signals instead of the printer hardware figuring it out for you : same
kind
of problem.
Some vendors did indeed supply an interface that took ASCII from the
computer and sent the necessary tilt-and-rotate signals out to the
Selectric.
Coincidentally, last weekend I was going through old magazines from 1979,
and found a pair of articles by Roland Perry in Practical Computing (the UK
magazine, January/February 1979) describing a Selectric interface.
According to the articles, there were three types: BCD, correspondence, and
BCD-converted-to-correspondence. All of them use tilt-and-rotate codes,
which vary according to the golfball type (BCD or correspondence) and the
keyboards differ as do the golfballs. The interface was pretty simple (8
SSI TTL ICs, 14 driver transistors) but the contacts had to be re-wired to
suit the terminal version. The driver code used a lookup table to convert
ASCII to interface signals, sent 8-bit-parallel from an I/O port on an
8080, with two handshake lines.
--
Pete Peter Turnbull
Dept. of Computer Science
University of York