Help restore a Flexowriter
jo...@cs.uiowa.edu
managed to make it work as an electric typewriter, by manually unlocking
the keyboard with a pocket knife. The keyboard feel is remarkably nice,
but the mechanism squeeks and is prone to jamming -- it obviously hasn't
seen any maintenance for well over a decade! Unfortunately, I can't
wake up the electromechanical (relay based) logic to make the machine
unlock its keyboard, and that means I can't make it talk to its paper
tape reader and punch. Furthermore, the machine has a "factory
programmed" plugboard with about 1000 contacts on it hidden inside,
and I have no idea what it does.
If anyone has any Flexowriter documentation, it would probably help me
handle the basic problem of oiling and otherwise maintaining the
mechanism. If anyone has specific documentation for the Singer
Programmatic Flexowriter 2201 and Singer Friden Selectadata box, I'd
be able to really figure out what's going on.
Having said this, here's what I've gleaned about the flexowriter from
the 2 hours I spent dusting the mechanism and peering into its guts:
1) The typewriter mechanism in a flexowriter is a classic "type bar"
mechanism, just like most old, pre-selectric, electric typewriters.
Pressing a key engages an eccentric cam against a spinning rubber
platten under the keyboard, and the cam slams the appropriate type
bar against the page to print the letter.
2) The binary code for each letter typed was encoded as follows: For
each type bar, a horizontal bar ran back under the carriage. The
horizontal bar would be either pulled forward or pushed backward by
its linkage to the type bar (half went each way). Each horizontal
bar had a loose (compression only) linkage to some subset of 15
crossbars, so that when the horizontal bar was pulled or pushed, it
would raise a subset of the crossbars. Each crossbar, when raised,
closed an electrical contact.
3) Given a binary code of a letter to print, a set of 7 bars running
side to side under the keyboard was shifted either left or right
(depending on whether the corresponding bit was zero or 1), and fingers
on the bars would part under exactly one key, allowing that key to
"self depress", engaging the typing mechansim.
4) The flexowriter used explicit shift codes, with a different code
for shift and unshift. Touch typing is difficult on this machine because
the only shift key is a locking shift and you've got to explicitly type
the unlock key.
5) The paper-tape reader and punch attached to the side of the flexowriter
are able to read and punch standard 8 channel 1" wide tape, but, given
the above observations about the shift codes, and given the non-ascii
glyphs on available on the keyboard, I suspect that there was only
limited compatability (if any) with the ASCII subset used by most 1960's
and later teletype-based computer systems. I found a snippet of
"Singer Flexowriter" brand paper tape inside the case, along with a few
other scraps and some chad.
6) The paper-tape reader and punch attached to the side of the flexowriter
are clearly designed to accept other media! It looks like they could
punch and read the right edge of a strip of adding machine tape, or even
the side edges of stationary (at about 90 characters per 11" tall page,
with about an inch of leader top and bottom).
7) Flexowriters are built like tanks. Heavy, all metal, and nothing
inexpensive inside. I suspect that a new flexowriter cost a more than a
bit more than a new ASR 33 teletype!
8) I find no evidence of a commutator or other parallel to serial
converter, but the flexowriter has a pair of big fat parallel interface
with about 50 pins each. In the 1950's, parallel to serial conversion
would have been expensive enough that you'd probably save money by using
the inherently parallel interface of a flexowriter instead of trying to
support the communications protocols used in the telegraph industry,
but by the 1960's, with inexpensive transistorized logic, you'd save
money by building extra logic in order to support an inexpensive Teletype
such as the ASR 33.
Doug Jones
jo...@cs.uiowa.edu
Joe Morris
>I now have a flexowriter sitting in the foyer of my house,
It makes a nice doorstop.
>4) The flexowriter used explicit shift codes, with a different code
>for shift and unshift. Touch typing is difficult on this machine because
>the only shift key is a locking shift and you've got to explicitly type
>the unlock key.
Those of us who used 5-level Baudot Teletype machines know all too well
the joys of forgetting to add the appropriate shift keystroke into
a transmission.
>5) The paper-tape reader and punch attached to the side of the flexowriter
>are able to read and punch standard 8 channel 1" wide tape, but, given
>the above observations about the shift codes, and given the non-ascii
>glyphs on available on the keyboard, I suspect that there was only
>limited compatability (if any) with the ASCII subset used by most 1960's
>and later teletype-based computer systems.
If it was an 8-level machine, it might be designed to use the same codes
as were employed on the PDP-1. <sound of creaking doors> I dug out my
old PDP-1 binder and found the cheat sheet for the FIO-DEC codes it used:
From MIT PDP-1 memo PDP-7 (memo number, not machine type!) dated 10/8/62
(condensed from a 3-page listing in the memo):
Characters Punch code (characters given as lowercase uppercase)
---------- ----------
aA thru iI 61 thru 71
jJ thru rR 41 thru 51
sS thru zZ 22 thru 31
0 -> 20 (right arrow)
1 " 01
2 ' 02
3 ~ 03
4 * 04 (shift character is "implies")
5 * 05 (shift character is "logical OR")
6 * 06 (shift character is "logical AND")
7 < 07
8 > 10
9 ^ 11 (shift character is uparrow)
([ and )] 57 and 55
-+ ,= /? 54, 33, 21
~| 56 (nonspacing; lowercase character is overstrike)
._ 40 (nonspacing; lowercase character is middle dot)
.x 73 (period and multiplication sign)
Lowercase 72
Uppercase 74
Space 00
Backspace 75
Tab 36
Carriage return 77
Tape feed 00 (representing blank tape without parity)
Red 35 (ribbon color select)
Black 34 (ditto)
Stop code 13 (stops tape reader when encountered)
Delete 100 (ignore this line on the tape)
The paper tape used one channel (octal 200) to produce odd parity, so
add octal 200 to each of the above (except DELETE) to generate odd parity.
Joe Morris / MITRE