12 Apr
2001
12 Apr
'01
1:59 a.m.
Well, that's certainly true. The EIA standard wouldn't spell out all those
signal name assignments if it didn't need them in the most general case.
However, in most cases, notably the popular DEC interfaces, it became common to
see just the three-wire hookup, probably because the DEC systems were tied to so
many terminals, often being front-enders, etc. That made for skinny cables that
didn't weigh as much nor did they take up as much space.
The first time I used a serial printer of my own, I found that the only way to
get around the all-too-common incompatibilities between printer handshake and
system handshake was to ignore the handshakes (i.e. turn them off) and make no
connections other than TxD, RxD, and Common.
more below ...
Dick
----- Original Message -----
From: "Tony Duell" <ard(a)p850ug1.demon.co.uk>
To: <classiccmp(a)classiccmp.org>
Sent: Wednesday, April 11, 2001 6:45 PM
Subject: Re: 50 pin SCSI to 50 pin centronics
> lives on a DA-15 connector. Back in those days,
RS232C spelled out the
signal
> levels and the signal names, and their functions,
though quite inadequately,
and
> it wasn't the EIA that authorized the use of
the DE9 for RS232-compatible
> communcation, though I don't know why, in view of the fact that only
DB25-pins
> 2,3, and 7 were required to get the job done.
Even today, lots of people
talk
Well, firstly, 2,3,7 are not always all you need. Synchronous interfaces
almost certainly need the clock signals (on 15, 17 and maybe 24?). You
might also need the modem control [1] lines, etc. There are many more
than 9 signals, so you can't always use a 9 pin connector.
And one reason for having a standard in the first place is so that (at
least in principle) any 2 devices meeting the standard can work together.
If some devices have a 9 pin connector and others have a 25 pin
connector, then you can't plug them together.
Yes, I know 'RS232 compatible' normally means a morning with schematics
and an afternoon with the breakout box working out how to link the darn
thing up. But that's because we're _all_ misusing the interface. It was
designed to link a teletype to a modem, and for that use, it seems to
work first time.
Sad but true ... The standard links the pins to the signal names, and sketchy
definitions of the signal functions, but doesn't even go so far as to define the
active state of the handshakes, thereby leaving much room for confusion.
[1] I have deliberately avoided calling them hardware handshake lines.
IIRC the standard is pretty clear about what you can do with them, and
hardware handshaking is not included. This makes sense in the original
application -- the RS232 link was between a terminal and a modem -- a
dumb modem with no internal data buffering or anything like that. The
device you wanted to tell to stop sending characters was the computer
connected to the modem at the other end of the line. The only way you
could do that was by sending signals -- characters -- over the telephone
line. You couldn't simply send a signal to the local modem, which is
what, of course, the control lines did. No, those lines were for things
like ensuring the modem was working, doing a half duplex turnaround, etc.
That's probably wise, and, in fact, wiser than most authors about this
subject.
Now, of course, we have a pretty good idea of how we expect those lines
to hehave -- as hardware handshake lines -- even though that is against
the standard. In fact when you find a device that uses them 'properly',
you might have a long day ahead of you making it work with just about any
other device out there.
Whenever I need a taste of humility, I get out my manuals for the HP 7585
plotter's serial interface and try to figure out how it's really supposed to
work. Though I've gotten it to work, I've never understood why it works the way
it appears to while the doc's say what they say.
-tony