Newsgroups: comp.dcom.modems
Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!casbah.acns.nwu.edu!nucsrl!ptownson
From: ptownson@delta.eecs.nwu.edu (Patrick A. Townson)
Subject: Troubles With Telebit T-1600
Message-ID: <1991Jun2.141605.5766@eecs.nwu.edu>
Sender: ptownson@eecs.nwu.edu (Patrick A. Townson)
Organization: EECS Department, Northwestern University
Date: Sun, 2 Jun 1991 14:16:05 GMT


I have quit using my US Robotics Courier 2400 modem and am now using a
Telebit T-1600 instead for my work (and play) on the net. It generally
works quite well, but one problem bugs me:

Configuration:

I use the 'tty' mode (AT&F0), with a couple differences from the
factory defaults. I have S254=0, although Configuration A and B are
identical. 

I use AT&D2, with the modem responding to DTR going up or down.

I set S51 (DTE interface speed) = 253 (autobaud, default to 38,400).

I set S61 (Local action on break) = 0, rather than its default of 1.
The default says pressing the break key will put the modem in command
mode. (This is in addition to use the of '+++' to do the same thing?)

My problem is I need to use the BREAK key occassionally to get out of
a trap when on line at eecs.nwu.edu.  There are times (in emacs, quite
often) when everything stops. I can't get out, nothing responds, and
the only solution is by hitting the BREAK key I at least get dumped
out back to the server ... so I wanted to keep the BREAK key for my
own use.

Since S61=0 (break is processed as defined in register 63), I then 
set S63=1, (send break immediatly).

S180=2, the default (error control is V.42 with detection).

S190=0, (Data Compression Enable) is turned off. The default is 1, to
enable in both directions.

    I use the modem almost exclusively in an interactive situation,
    typing at the terminal. I rarely download/upload big files. Mainly
    I wanted the screen refreshes to be faster ... and they are much
    faster than with the old 2400 baud unit.

    The only thing Data Compression does is causes a 'jerky' effect as
    the stuff being transmitted is put in packets, etc ... I did not 
    need it so I turned it off.

Otherwise, the registers are all set to the factory default.

The problem:  using three plusses to go to command mode works okay in
getting me there, but I can't get back on line.

Using the BREAK key to do this (by setting S61 back to the 1 default)
works okay ... I go to command mode, and ATO puts me back on line.

Using the three plusses gets me into command mode, and ATO *says* I am
back on line (produces the CONNECT 9600 message), but nothing actually
gets from the DTE to the modem and on to the distant modem at that
point.

I tried ATO1, which the manual says 'retrains equalizer before going
back on line'. ATO is also possible to go back on line without
retraining. 

Once I have 'three plussed' into command mode and did ATO or ATO1 to
go back on line, my only option at that point is to disconnect
entirely and dial back in. 

I could see this happening if I changed some registers while in
command mode, or otherwise got the modem confused, but I've tried this
doing nothing other than 'three plussing' into command mode and
immediatly issuing the ATO or ATO1 to get back out ... and it still
fails to reconnect.

Finally, after plussing => command mode => ATO => back on line, but
with an unresponsive modem and typing blind (nothing at that point
comes back to the screen), if I three plus again, the three plusses do
not show up on the screen, but the modem does act on them by putting
me back in command mode again. 

We wondered: does this problem only occur at 9600 baud?

    Logging in at lower speeds (1200/2400 baud) on the same unix
    machine got the same results ... three plusses into command
    mode, then ATO back on line, but typing blind and no response
    from the modem except for the entry of three more plusses.

    Logging in at lower speeds to MCI Mail / AT&T Mail and a couple
    local BBS' did **not** produce this problem. When on those sites,
    the three plus => command mode => ATO back on line worked fine.
    Unfortunatly, the mail services and the local BBS' here do not
    support 9600 baud (that I know of) so I could not test the
    higher speed there.


Can anyone suggest a solution?  

    Why will BREAK => command mode => ATO (ATO1) => back on line work
    fine at any speed on the T-1600, but the three plusses exit to
    command mode fail to go satisfactorily back on line, causing blind
    typing and no response?

    Why does this occur when calling this unix site, but not when
    calling MCI Mail (for example)?  


Thanks for answers.


Patrick Townson