[inforoots] TTY history (was: Restoring the WISC)

Mike Albaugh albaugh at perilin.com
Wed Feb 2 09:42:03 PST 2005

> Actually we cold use an array of ccd cams each set to trip on a different
> bit and thus catch the actual motion of the mechanics.

 Can anyone comment on the feasibility of this? That is, are there
cameras for whih an external signal can "trigger" them with a small
enough, and constant enough latency to make this sort of
"Muybridge for the 21st century" approach work? Are the cameras
Texx is thinking of _much_ better than typical "snapshot grade"
digital cameras, like my Sony PSC-8? I ask because that one has
what seems like a long and variable lag between pressing the shutter
button and "taking the picture", and most of the cameras I've seen
that support "movies" at all do it at fairly low frame-rates, from which
I infer that we would still need a strobe, if for no other reason than
avoiding as much motion-blur as possible.

> Hand cranking the motor drive is a good exiibit Ive used in the past to 
> demonstrate MY model15 (TT-7-FG)

In my experience, that works to a point, but some parts (especially the
actual printing) depend on dynamics that are just not present at low
That's why I did the strobe hack.

> We have several baudot machines in storage of various models.
> Some day when we get to next phase,  Id like to work with curators
> on a TTY exhibit.

 Sounds good to me, although (since we're here on a list of folks who
might know) what I'd _really_ like to find is more info on a machine I
in a historical column in QST, back in the 60's. It was a keyboard Morse
sender, all mechanical, and what was intersting (and relevant to
is that the keystrokes were stored in a FIFO, to allow the operator to
in a natural rhythm while the machine took a variable amount of time to
send each character. Compare and contrast to the need to develop
that Green-Keys Beat when using a Model 15. :-)

> Im preparing to place a model28 back into full time service at my MARS station.

congrats. Love that machime. First use of mechanical pipelining I ever

> Regarding UARTS.
> The model14 reperf 
> (Why the hell they call it a reperf, Ill never know...
> The tape was ALREADY PERFED and all this thing does is read the tape !
> If you have both the model 14 reader AND the model 14 punch THEN the WHOLE set 
> up might be a reperf...)

 Some explanation. There were at least three machines called "Model 14".

1) The Model 14TD (Transmitting distributor") is what most people think
    It is essentially a tape-reader with parallel->serial convertor
   a big rotary-switch). Sometime seen alone but mostly as one part of
   Model 19 (what later might have been called a 15ASR :-)

2) The Model 14 Strip Printer. This is the machine that Western Union
    a lot. It had a type-basket mechanism much like the Model 15, but
    basket was fixed to the base-plate, and the type-bars struck down to
    a small wheel-shaped platen. The output was a strip of paper which
    torn at appropriate line-breaks and pasted down on a pre-printed
    to make the sort of Telegram that people imagine when they hear that
    word, if they are old enough to have the mental image at all, or
    a lot of old movies :-) I suspect that the fact that WU used a
    different timing and minor variation on character code meant that
    of these got through the surplus market and into the hands of  hams
    and the deaf.

3) The Model 14 Printing Reperforator. This was very similar to the
    Printer, but added a punch near (but not exactly _at_ ) the platen.
    The (serially) received characters were both printed and punched.
     To make this work, the printed character was offset from the
     version (6 characacters, if I recall correctly, but it's been a
_long_ time)
     and the tape was "chadless". That is, the holes were not completely
     punched through, but left attached by a small bit of hinge. This
     had the benefit of not requiring a chad-box and someone to empty
     It also implied that optical readers could not be used. You needed
     read these tapes with a pin-sensing reader like the 14TD, or, more
     likely, the readers that were used in the "coke machine" officially
     known as the "Torn Tape Message Relay". Those readers were more
     mechanical that the 14TD, using the 'Cam and Seahorses' mechanism
     similar to the keyboard serializer on the 14/15/26.

> Anyway, it was capable of connverting parallel from tape holes to serial.
> Ive NEVER seen a mechanical device that converted serial too parallel 
> electrical.

   From various articles, I've gathered that the Model 12 did this. That
there was a "distributor" like that of the 14TD, but started by the
bit", and it was used to drive in sequence each of the five magnets
that controlled the key-basket selectors. I've never seen a Model 12,
partly because they were old at the time I got into this, and partly
this "mostly electrical" approach made them too (electrically) noisy
to work well for RTTY. The big advance in the 14/15 line was the all-
mechanical selector mechanism, which survived through the 37.

>  If someone took a  model 15 assembly and put switches on the ends  
> of the  select bars, you could do it but I never saw it done.

 I never saw _that_ done, either, but I did use a 14reperf. Short
description/picture at www.perilin.com/albaugh/coke_machine.html.
> 6SN7s ?  Isnt that overkill for a UUART ?

Why? If I recall correctly, the 6SN7 was the dual-triode of choice for
octal-base tubes. Later tube computers used the 12AT7/12AX7 or
"computer grade" rough equivalents, but if I wanted to build octal-
based logic (no Hex-jokes, please), I'd certainly use them. What do
you find wrong with that?

> Im betting that the keyboard used a separate circuit from the print loop.
> Much easier to intereface that way.
> The model 15 was wired that way and if you wanted them in the same loop you
> just tied the jacks together.

 Right. Most (all?) Teletype gear was arranged that way, except the
paper-tape punches on the Model-19, which got its data directly, and
in parallel, from the ends of the keyboard encoder slides, and on later
"ASR" machines, which got their data directly, and in parallel, from
the print selectors. I used to have a "Singer Conversational Terminal"
(late-model Flexowriter) which was inherently half-duplex (keyboard
hard-connected, mechanically, to the print mechanism) and I believe
the IBM-executive conversions were also that way, but not TTYs.
On the other hand, Paul Pierce has already said that WISC originally
had a Flexowriter, so...

> I suppose Ill have to find time and visit the archives and look up the 
> schematics to the beast.



More information about the inforoots mailing list