JP>>> The only problem I've seen with the Roadsters is their
annoying inability 
JP>>> to handshake. Upon making a call, they seem to get stuck in
a continuous 
JP>>> loop, trying desperately to train.
LE>> I've seen this with SEVERAL Rockwell modems after V.34+ upgrades. The 
LE>> problem seems to occur ONLY under the following conditions :
LE>> a) The V.34+ Rockwell modem is originating
LE>> b) The other end is any V.34+
LE>> c) The connection "was going to be" 33600bps
RB> It must be in v.34 (28800) generally.

The examples you gave "once", "occasionally" etc are
explainable as random acts of dog. If the conditions I listed above are
met, the problem is 100% reproducible, indicating a systematic flaw
somewhere.

LE>> hangs at the end of the handshake, always.
RB> Problems with v.42 detection maybe...
RB> Try forcing v.42 without trying to detect it.

It happens during low-level protocol init, before EC/DC negotiation takes
place. The modem speakers are still on. I think the originating modem has
hit an unrecoverable error and is waiting for the other end to start the
neg sequence again from the beginning. The answering modem is stuck in the
very last phase of the transport negotiation.

-- Lewin A.R.W. Edwards [Team OS/2]  Tel 0412809805 * http://www.zws.com/
@EOT:

--- MsgedSQ/2 3.35