[net.dcom] Noisy {{~r lines affecting 212A { modems

vanam@pttesac.UUCP (Marnix van Ammers) (01/14/85)

The following was originally posted to net.misc .  I now know
it should have gone to met.dcom .  Sorry.

Does anyone know why a noisy line on a 212A data
connection results in one or the other end seeing
the RUBOUT or DELETE (127, 0177, Hex-EF) or the left
brace (123, 0173, Hex-EC)?

I've also seen the tilde (126, 0176, Hex-EE) followed by
the lower case 'r' as well.  At work I had "~r" a lot.
Lately at home I get the RUBOUT (which creates havoc with
my editing and will soon drive me to suicide!).

There are other problems I run into as well with
the 212A (official Bell version).  Occassionally
I make a connection and the "MC" lamp on the data
set doesn't go out and I can't send or receive over
the connection although the connection does stay up.
Both 212A's were supposed to be in factory configuration
although I didn't verify this myself.  I suspect this
is a "level" problem.  People close to the destination seem
to be able to work with these modems.

honey@down.FUN (code 101) (01/15/85)

i sometimes get noisy lines, so i share your frustration.  i have tried
changing my interrupt character, to no avail.  i conclude that DEL is
not actually being seen, but that the uart is getting a framing error,
which turns into a break, which turns into your interrupt character.

no i do not have an answer (other than griping to the phone company).
	peter

jcp@brl-tgr.ARPA (Joe Pistritto <jcp>) (01/17/85)

	The errors you get with the 212A modem are almost all burst
errors, caused by the scrambler/descrambler circuit in the modem.
When any bit is damaged, because of the scrambled nature of the
signal, (which is scrambled in 'groups' of bits (I think 4)), all
of the bits in the group are damaged, typically turning into ones,
hence the high probability of characters with lots of ones in them
(RUBOUT 0177, Tilde 0176, } 0175.  This is one of the worst features
of the 212 design, and renders it practically useless for interactive
lines except where line quality is better than average.  Try VADIC
3400 protocol modems if you can, they don't have this problem (or
many others for that matter...)

						-JCP-
PS: I don't work for VADIC, I just like their stuff.

rjk@mgweed.UUCP (Randy King) (01/17/85)

Ugh, this topic rears it's ugly head again.  I, too, am still living
with the problem, knowing full well what it is and how to fix it.  No
one at Illinoisy Bell really gives a rats arss about my problem.

It is a CO synchronization problem.  Over a T1 digital trunk, if the
offices are not sync'ed, a "digital slip" occurs periodically which is
the result of one end re-synchronizing.  This causes a PSK error that
hoses one direction of the 1200 baud signal.  Fortunately for me (??)
that error is only "seen" by my terminal, and not the host.  Really screws
up vi and other terminfo-based programs.

						Randy King
						AT&T-CP@MG
						ihnp4!mgweed!rjk

mjl@ritcv.UUCP (Mike Lutz) (01/18/85)

In 418@down.FUN :

> i sometimes get noisy lines, so i share your frustration.  i have tried
> changing my interrupt character, to no avail.  i conclude that DEL is
> not actually being seen, but that the uart is getting a framing error,
> which turns into a break, which turns into your interrupt character.

Hmm.  On the noisy line that I sometimes have to use, I almost always
see a DEL/{ pair sent to the remote system.  I set the terminal in 8-bit
no parity mode and ran 'cat' to pick up some of these alpha particles.
Sure enough, the DEL was a real DEL with all 1 bits -- apparently noise
forced a 0 (start bit) and recovered immediately to the default 1 state.

This could be fixed by setting the terminal to odd parity, 7 bits, but
then you only cut out the DEL - unfortunately, the pattern we see for
the { would pass an odd parity test.  We've thought about implementing
a SPACE parity option (parity always 0), as our junk characters almost
always have the parity bit set to '1'.  Anyone want to hack on this?
-- 
Mike Lutz	Rochester Institute of Technology, Rochester NY
UUCP:		{allegra,seismo}!rochester!ritcv!mjl
CSNET:		mjl%rit@csnet-relay.ARPA

mroddy@enmasse.UUCP (Mark Roddy) (01/18/85)

[ 212A line noise]

	It's a 212A feature- poor filtering capabilities. However there
	are two things to do - 
		1. hang up and call again, your
		next connection might be better. Also
		in this category is have the phone company
		check out both sides of the connection;
		Technically the phone company is supposed
		to provide data capability on the line if you use
		a certified 212A modem. I have actually gotten
		results from this approach- it turned out that
		one of our lines had a second live connection
		on it, once disconnected data quality was restored.

		2. use parity- most everybody configures
		for no parity, consequently even this limited
		error checking is circumvented.

george@idis.UUCP (01/20/85)

I use a 300 Baud acoustic couple modem.
I seem to often get a noisy line.
In its more mild forms the noise looks like deletes.
I tend to notice this quickly since my default interrupt
character is delete and ctlecho is set.
Calling back does not always eliminate the problem.

At one time I tried fiddling with parity,
but this seemed to be too much of a pain.

I now have a (UNIX) shell script called "noisy"
that consists of the below line.

	stty intr '^C' start '^?'

If I notice an unexpected delete, I run the "noisy" command.
Deletes are ignored.
(At 300 Baud, the terminal doesn't need start/stop.)

This does not fix the problem, but it makes it
tolerable enough to read news.

	George Rosenberg

vanam@pttesac.UUCP (Marnix van Ammers) (01/20/85)

I've received about 5 mails re: my noisy {{ ~r 212A posting.


The consensus is that most of the noise I experience is
caused by lack of synchronization of digital carrier facilities.
Due to the way the 212A encodes data, the "time slips" in digital
carrier facilities usually cause 1 or 2 bit errors.  If the error
occurs while the facilities are idle (marking) then the noise
will usually result in characters with mostly 1's.

Now the thing to find out is (and I will try):  Do the
digital carrier maintenance forces know about unsynchronized
systems?  Do they get alarms?  Are they alerted?  Is it easy
to fix?

If you know the answers, I'd like to hear from you.  I'll post
whatever I learn.  Thanks to everyone who responded.

-- 

Marnix (first name -- pronounced as marnix)
van Ammers (last name -- 'v','a','n',' ','A','m','m','e','r','s')