[net.dcom] Phone line blues

mosier@iuvax.UUCP (Steve Mosier) (03/05/86)

In September, the local phone company (Indiana Bell) changed over to a
digital switching system.  Since that time, trying to use a modem has
met with less than desirable results.  After several days of dealing
with the local CO, a tech type came to the university and installed some
test equipment on the lines and came up with error rates that exceeded
his wildest expectations.  He finally concluded that the "lines were
out of phase"  His efforts to correct the problems seems to have
succeeded, somewhat.  However, the preformance is far from what it was
previous to the switch over.  The phone people seem to take the attitude
that "if you can hear and talk to the other person, everything is ok"
For example, I can login on to what appears to be a clean line and 
everything will seem fine, then I get a stream of 30 or 40 garbage characters.
I can go along for several minutes, then I get hit with another stream.
Dialing out for UUCP traffic seems to meet with the same results.
Our resident engineer maintains that, although you can hear and talk, 
the lines are probably much out of spec since a modem that is designed to
operate over a voice grade line simply can't.  I've called the phone co
several times now, with no results.

One note on the above, the phone lines do not have any audible clicks
or pops, so I'm at a loss to explain why they give so much garbage.  The
garbage is not local echo i.e. it will appear as input.  Another curious
factor; folks within the city have little or no problem dialing in, but
people on the fringe areas and from out of town get hit the hardest.
We have checked and double checked our dialins and dialouts, so the ball 
is definitively in their court.  How does one deal with this?

-steve  mosier@indiana.csnet or {cbosgd,ihnp4,seismo}!iuvax!mosie

smithrd@gc49.UUCP (Randy D. Smith) (03/05/86)

In article <223@iuvax.UUCP> mosier@iuvax.UUCP (Steve Mosier) writes:
>In September, the local phone company (Indiana Bell) changed over to a
>digital switching system.  Since that time, trying to use a modem has
>met with less than desirable results...
>...He finally concluded that the "lines were
>out of phase"  His efforts to correct the problems seems to have
>succeeded, somewhat.  However, the preformance is far from what it was
>previous to the switch over.  The phone people seem to take the attitude
>that "if you can hear and talk to the other person, everything is ok"...

Do you have voice grade lines?  Or data lines??  I expect that each
is fully specified, and that the phone company there should know
how to ensure that the specs are met.

If you are dealing with voice grade lines, then they're probably "balanced"
lines.  They work for voice, and the circuitry that makes them "balanced"
adds to the line's quality as a voice line.  Data lines are USUALLY
unbalanced; that circuitry that makes the line balanced, and improves
the quality of the voice line, introduces the phase differences that
interferes with the line's usefulness for data transfer.

>...Our resident engineer maintains that, although you can hear and talk, 
>the lines are probably much out of spec since a modem that is designed to
>operate over a voice grade line simply can't.  I've called the phone co
>several times now, with no results...
>-steve  mosier@indiana.csnet or {cbosgd,ihnp4,seismo}!iuvax!mosie

When faced with deciding whether an operating companies lines are
"out of spec" or whether a manufacturer has designed for "normal
operating conditions", I'd probably assume the latter.  People
have used voice lines for data for a long time, and don't think
anything about it until it stops working.  Then they find out that
there is no guarantee, unless they spring for the data line.

I ran into an equally frustrating experience a couple of years ago.
After a year of error free communication between the CA bay area
and NJ, I moved back to NJ within 5 miles of the computer I was
talking to, and with the move, I lost my ability to communicate
with the machine.  Five miles, versus 3000, and in the heart of
"phone company land"!  (Actually, NJ is owned by Prudential; AT&T
just leases it :-) )  I had a data line installed, with no improvement.
Fortunately, after a couple of months of frustration, the union
workers went on strike, and the operating companies had to send
engineers out for simple requests like mine.  The engineer who
worked on my line said it was one of the strangest things, but
that he could get it to work.  He said he had to run a balanced line
from the central office to a point about half way to my location,
then run an unbalanced line the rest of the way.  All I knew was
that it worked (until the first thunderstorm, but that's another story).

Note: I work for AT&T, but I definitely don't represent "the phone
company".  I'm posting only to add a little light to the subject.
I post from the perspective of someone who has run into similar problems.
I'm a software-type, not hardware, so take it easy on my (perhaps incorrect)
usage of terminology.
-- 
				Randy D. Smith	(919) 279-5312
			AT&T Technologies, Guilford Center, NC
			....!{ihnp4,burl}!{gc49,gc3ba}!smithrd

klr@hadron.UUCP (03/07/86)

	I have similar horror stories to tell.  I have no problem
	communicating at 2400 baud with FIDO nodes in Holland, England
	and California, but I am forced to use 300 baud to communicate
	with nodes that are no more than 10 miles from me.  Other nodes
	in the local area can not be connected to at all.  The carrier
	that the remote is generating is so distorted by the lines,
	that my modem does not recognise it as a an answering carrier.
	And the situation has been getting worse with the the advent of
	the "dial 1" disease.

	Kurt Reisler
	SYSOP on the following:
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
                     _
The World's First   /  \		Whose dog is that?
   BBS Network     /|oo \		I don't know, but the disk is yours.
   * FidoNet *    (_|  /_)
                   _`@/_ \    _	        The Bear's Den FIDO 109/74
                  |     | \   \\		(703) 671-0598
                  | (*) |  \   ))	Wash-A-RUG     FIDO 109/483
     ______       |__U__| /  \//		(703) 359-6549
    / Fido \       _//|| _\   /
   (________)     (_/(_|(____/ (jm)
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

brian@sdcsvax.UUCP (Brian Kantor) (03/09/86)

We had a similar problem where when the phone co installed a new switch
- some exchanges could work without errors, others were completely
unusable.  I'm told that it was that the digital trunks were not
synchronized, and that after some clock rate tweaking things got a lot
better.

The bursty sort of errors you describe sound like a sync problem - where
the lines loose sync and have to recover it.  Normally this is nearly
inaudible, but it tears the hell out of synchonous signals like modems
and such.

How did we cure it?  By having lots of trouble calls placed by the
system administrators of the computers affected.  After a week or two
the PhoneCo got somebody (I think I heard it was an engineer from the
switch manufacturer) who knew what he was doing and got things
corrected.

It works pretty well now.  Except that we had to add a delay between
dialling 9 and the rest of a number, because the new improved switch
can't handle 91234567 as quickly as the old one could.  Sigh.

Oh yeah, voice always sounded good.  And 300 baud modems worked just
fine.
	- Brian