F8 and a GPS -OT (HTML for drawing)
pyevet at aapt.net.au
Wed Jul 9 23:49:41 UTC 2008
On Wed, 2008-07-09 at 00:57 -0400, Gene Heskett wrote:
> On Wednesday 09 July 2008, Les wrote:
> >Gene and others will correct whatever I get wrong...
> > The original specification was for a serial interface to multiple
> >devices. If you look at the entries below 0x20 on the ASCII table, you
> >can gather such gems as DC1 and SEND and some others. These were meant
> >to control multiple devices on a single serial port. This is part of
> >the software specification. The hardware specification implemented two
> >serial busses in a single connector. Although most hardware never
> >implemented the second bus, instead treating the one as bidirectional,
> >which reduced the effectiveness of RS232. Also most software folks
> >outside of IBM and UNIVAC seldom implemented the full control spectrum.
> > The line drivers were supposed to be cabable of 20ma loop
> >current. This meant that the serial loop could pass through a chain of
> >devices, each of which could receive and transmit data.
> >You could select which device(s) were active by chosing device control
> >codes which would act on specific hardware in the receiving device. Its
> >primary use was for communications, driving data to modems and receiving
> >data from modems.
> > Generally most folks would ground the shield, which was sort of an
> >error. What should happen is that the shield is the return path for the
> >transmitted signal. from the DTE. Signal ground is the return path for
> >the signal originating at the DCE. Thus properly implemented the two
> >sides have different ground references. At the DCE (modem or printer
> >for example) the signal ground carries the return path for the signals
> >going to the DTE. the shield carries the signal return to the DCE.
> >Thid implied that who ever was working on the line had to remember which
> >line he was on. 200' of line in a high RF environment or high loss
> >environment could build up quite a charge if it were not dealt with
> >properly. Inside the equipment, the receiving block at the DCE would
> >tie its input circuits between shield and a floating 5v signal. It was
> >never really floating, as there would be a high value resistor between
> >the shield and signal ground. The transmitting block of the DCE would
> >be tied between DCE +5 and signal ground.
> > At the DTE end its transmitting block would be tied to +5 and shield
> >to ground, with signal ground connected by a high value resistor. Its
> >receiving block would be between a floating 5V and signal ground.
> > NOW all that is the theory. In practice shield ended up being
> >connected to case grounds on both ends, and signal ground was used as
> >the reference. This meant that you had some signal degradation on
> >simultaneous transmit and receive, resulting in the buss being used
> >alternately for send and receive. Also most manufacturers didn't want
> >floating supplies, so that ended. Finally they balked at the additional
> >$3.00 or so for the line driver bits, and used normal 7400 or DTL logic
> >for the lines. the result was lower speeds, shorter distances, loss of
> >noise immunity, and too many unused pins on the connector. In came the
> >DB9, and a host of other hardware interfaces, out went the lines
> >available for ringing (the anonymous test lines), and so went the story
> >of slow, limited line length RS232.
> > So basically all you need is the pinout in the bottom connector of
> >the drawing (DB-9) or one of the other interfaces, short leads
> >(generally <6') and you should be all set. There isn't much that can go
> >wrong other than you may need a null modem connector to transpose the
> >two serial wires. They are available all over the place, or you can
> >simply open up the offending connector and swap them there.
> >Les H
> > The original interface was a 25 pin device:
> >from: http://www.camiresearch.com/Data_Com_Basics/RS232_standard.html
> And that is a better description than I could ever have written. I wasn't aware
> that some of the odd signals were actually a second data channel. I also do
> not think any of that is covered in "Understanding Serial Communications" or
> whatever it was that the Shack was selling as The Bible of rs-232 20 years ago,
> and which I have a copy of lolling around here someplace.
> Unforch, the hardware side is only half of the story, the handshaking to assure
> good data is the other half, and that can get fairly complex too as it appears
> there is more than 1 interpretation of the '7-wire' protocol now so widely used
> with the db9 connector cuz even that leaves 1 empty pin. The RI is rarely
> connected or used, although some modems may have exerted it back in the dim
> reaches of time 2 decades ago.
> Cheers, Gene
> "There are four boxes to be used in defense of liberty:
> soap, ballot, jury, and ammo. Please use in that order."
> -Ed Howdershelt (Author)
> Fishbowl, n.:
> A glass-enclosed isolation cell where newly promoted managers are
> kept for observation.
This has jogged the memory a bit. A friend has a book by Scott Mueller
that has a bit on serial I/O. With all this and google I should be able
to get started. That rainy day is coming...
More information about the fedora-list