http://hackaday.com/2016/05/29/dragging-teletypes-into-the-21st-century

[Christian Gauger-Cosgrove]

On 29 May 2016 at 20:15, js at cimmeri.com <js at cimmeri.com> wrote:
> Tangent: is it true as written that *all*   "teletypes speak 5-bit ITA2 code
> <https://en.wikipedia.org/wiki/Baudot_code#ITA2>"?
>
As Ed, Jon, and Chuck said; no not *all* teletypes speak 5-bit code.
The "most common" machines in the classic computing community of the
Model 33 family (ASR and KSR) speak 7-bit ASCII 1963 with a parity bit
(either marking parity, or even parity; usually).

The 5-bit equivalent to the Model 33 is the Model 32, as Chuck
mentioned. Similarly, there are other ASCII machines, the Model 35
family, for one is absolutely beautiful; it's based on the mechanisms
of the older (and tank-like) Model 28 family. In fact if I recall what
was said on the Greenkeys list, the Model 35 parts all have the same
names as their corresponding Model 28 parts, except with an '8' in the
name. But ask an actual expert.


Also, just to be slightly nit-picky, many 5-bit code speaking Teletype
machines made in/for the US speak the similar, but not quite the same
USTTY code. It only differs in the figure shift, where the bell and
apostrophe characters swap places (bell is on J in the standard, S in
USTTY), WRU (figure shift of D) is replaced with a normal printing
character (dollar sign), and the "national use" characters of F, G,
and H are defined as the excalamation mark, ampersand, and number
sign. The remaining figure shift characters are completely compatible
(and the letter shift doesn't differ at all).

Hence why in the ACPs that deal with teletype communications the
acceptable characters for use include only those which are held common
between the two versions of ITA2 and USTTY. And also why the bell code
for Flash precedence messages is <FIGS>JJJJJSSSSS<LTRS> (so that no
matter if the message ends up in Europe or the US, the machine's bell
will ring). (See: ACP-127(G), paragraph 137.e.)


Also, being I'm a process engineering student: The article makes it
seem like current loop signalling is a weird dead interface standard;
it's not. Current loop isn't that confusing a thing to deal with. We
still use it for whenever we need to have electrical signalling that
is more noise resistant or goes a longer distance than voltage level
signalling. Problem is that current loop converters cost way too damn
much. Also, unlike the generic "RS-232C on one end, 20mA current loop
on the other" converter boxes that are dumb as a bag of hammers, the
board in question has some active electronics for conversion "stuff";
which makes the board infinitely more cool.


Cheers,
Christian
-- 
Christian M. Gauger-Cosgrove
STCKON08DS0
Contact information available upon request.