[GreenKeys] IBM 2741

[Dave Emery]

On Thu, Jun 16, 2005 at 08:51:39AM -0400, Tim McNerney wrote:
> Since we are on the topic of IBM printers, etc...
> Does anyone have any memories to share on the IBM 2741
> printing terminal?  This was the data version of the IBM Selectric.
> I think it was half-duplex, because it had the ability to physically
> lock the keyboard when the mainframe was unable to listen to
> the user's typed characters.

	I SURE do.   I'm not sure I ever saw one in the flesh back then,
but back in 1976 I was given the task of emulating one on a DG Eclipse
computer.   Basically someone handed me the IBM programmer manual for the
thing and said "make the DG machine talk to an IBM front end terminal
controller as if it was a 2741".

	I think the idea was that some insurance company computer would
import data from the DG machine thinking it was typed into a real
2741...

	In any case it did have a fairly complex protocol (including
locking the keyboard) and talked over a multi-drop 4 wire VF data line
using a 202 (similar to ham AFSK packet on VHF) FSK modem at 1200 baud. 
It got polled by the IBM front end processor and turned on the carrier
on the modem and transmitted a brief spurt of response or a longer burst
that contained one typed in line from the keyboard (or maybe also a
field in a form, I forget the details).   And the host spat back a
buffer full of characters to print (forget how big, but a couple of
lines of text at least, maybe more).   There was a mechanism to say "the
buffer is full" to pace host input to be printed.

	The burst of input data (and the output from the IBM host) was
checksummed and if the 2741 or the host missed a bit there was a
mechanism for a retransmit (and an error timeout).

	And needless to say it was addressable on the multi-drop line, I
forget but perhaps 64 or 128 of them could hang on one multi-drop 
connection, each at a different address.

	Off hand I forget whether it was EBCDIC or ASCII - I vaguely
think it wasn't standard ASCII at all and I had to do the translation
from its code set to and from ASCII.

	All of this was a bit of fun for a young kid like me back then,
and it was the very first time I ever got to meet a data scope (protocol
analyzer) and actually use it.    Think I used one of the first Spectron
DataScopes to debug my code talking to the host.

	And like many such "specials" the contract with the insurance
company got canceled before the thing ever got actually deployed in the
field.   But my emulation worked fine and passed its tests with the host
and would login and do all the right stuff...

	The funny thing was it was only maybe 20 years later that I
actually saw a real live 2741, all the testing had been over a data line
to the data center rather than from a site that actually had a 2741.

-- 
   Dave Emery N1PRE,  die at dieconsulting.com  DIE Consulting, Weston, Mass 02493