On Feb 03 05:32 96, Arthur Marsh of 3:800/812 wrote:

AM> On Thu 01 Feb at 18:33 Dave Hatch (3:711/808{at}fidonet) wrote to Paul 
AM> Edwards:

DH>> It's not happening to just USR's.   The USR diagnostics are 
DH>> good enough to rub your nose in it - but once knowing it was 
DH>> there and how to look for it, I got exactly the same thing 
DH>> with a pair of Hayes 28800 Optima vFC's.

DH>> Jury's still out on what the dickens it is.  That it exists 
DH>> is no longer in doubt.

AM> Brian Peacock at Telstra INOC has told me that it should be fixed and 
AM> working by Tuesday, if you're referring to international calls in the above 
AM> problem case.

Nope.  Local.  Even within the same exchange.

AM> I wouldn't blame Telstra at this stage either. One of their suppliers 
AM> provided non G.165 compliant echo cancellers without documenting the 
AM> deviation from compliance )-:.

AM> The short answer is that V.34 ensures that at no time is there no energy on 
AM> the phone line during V.34 startup and operation (although line probing 
AM> depends on one end of the link being silent so that the line's echo 
AM> characteristics can be determined by the modem at the other end of the 
AM> link). G.165 compliant echo cancellers should be disabled on reception of 
AM> 2100 Hz tones with phase reversals (which V.32, V.32bis and V.34 startup 
AM> provide), and not kick in again unless there is no energy in *either* 
AM> direction.

The symptoms are a sudden and severe asymmetry - had a hard lock on 28800
one way, 16800 the other today, for instance, two different lines through
the same exchange.  That one happened to maintain connection, although it
retrained every two or three minutes.

Normally the fault won't permit a successful retrain.  It's our old friend
the "sudden disconnect" caught up with.  IMHO, it's been with us
for years, and was the primary cause of all the problems getting first VFC,
and now V34 modems stable and usable.

Characteristics - occurs on high quality lines, with 31200 happily running,
good S/N, excellent line phase and amplitude characteristics - perfect. 
Wham.

One direction or other will suddenly begin frantically downshifting - in
one case, I saw it all the way down to 2400.  In most cases, the effort
fails, and the call disconnects.  If you hear it with the speaker on, the
sounds are rather astonishing - something VERY funny is going on.  It's not
a crossed call, however - no other channel comes through.

It is sensitive to traffic - and especially doesn't like late afternoons.  ????
Late night is usually immune to it.  When the fault is at its worst, a
connect will last no more than 2 to 3 minutes average.

Final interesting bit - if you set the call up at a low enough baud rate,
it doesn't happen.  9600 is reasonably immune.  (But if the fault is
triggered, 9600 isn't gonna work any more.)

It's not all that sensitive to modem type.  My VFC's, the USR Courier, and
a Hayes V34 all do much the same thing, when the plague is on.

Regards,
Dave Hatch.

--- Msgedsq/2 3.20