bruce@think.com (Bruce Walker) (11/09/90)
Since upgrading our dialin modems to the rackmount version of the Telebit
T2500 and handing out a bunch of T1500s for dialups, we have had lots of
problems with connections dropping in V.32 mode.
The problem seems to happen when MNP is turned on ("CONNECT 9600/REL") or
when V.42bis is turned on ("CONNECT 9600/LAPM"), and it is clearly related
to line noise: most of the problems are with long-distance connections or
known-to-be noisy lines, and in situations where connecting without MNP or
LAPM is too noisy for use.
The symptom is a long pause, then a "CONNECT 9600/REL" message (or
occasionally a "CONNECT 9600"), regardless of whether the connection began
with LAPM or MNP. After a few more seconds, carrier is always dropped.
Sounds as if it is trying to renegotiate in the middle of a session.
This has also happened using FastComm V.32 modems as dialups, so it is not
a T1500-specific problem. It is also not specific to one or two specific
dialup modems either; I have a bank of 12, and it happens regardless of
which one I use. PEP never has a problem like this on the same phone
lines.
Telebit technical support had nothing useful to say, other than they didn't
have a record of anyone else reporting it (which I find hard to believe)
and that it is allowed to renegotiate MNP off if there are too many errors
during the session, hence the "CONNECT 9600" message. I haven't looked at
the protocols, but that must be bogus information, since it completely
defeats the purpose of error correction.
My questions:
1) Is this a Telebit implementation problem or a fundamental
problem in the protocols involved?
2) Has anyone else seen it on Telebit or other hardware?
3) What can I do about it?
I am holding off purchases of additional modems until we can do something
about it. I'll summarize responses.
--
--Bruce Walker
Thinking Machines Corporation, Cambridge, MA
bruce@think.com; +1 617 876 1111mam@mamnix.hamm.sub.org (Michael A. Meiszl) (11/15/90)
bruce@think.com (Bruce Walker) writes: > 1) Is this a Telebit implementation problem or a fundamental > problem in the protocols involved? > 2) Has anyone else seen it on Telebit or other hardware? > 3) What can I do about it? I have the same difficulties with my two TB2500. V42 does a better job as MNP, but both of them resign after a hit by line noise. I have contact with a manufacturer of V.32 modems here in germany and their devolper tracked the problem quite down. He found that TBs generate a bogus MNP/LAPM header on resynch. They just send a byte too much, so the other modem can't resynch. This happens too often, the line is dropped by the other modem after getting too much errors. The new Roms (GF7.00) behave a bit better, V42 can be resumed, MNP is still bad. I personnally gave up on this error. PEP works fine and the next TB won't have V.32 anymore. I save my money for other nice things.. cu mam -- Michael A. Meiszl, Tomsonstr. 25, 4700 Hamm 1 Internet: mam@hamm.sub.org Voice: 02381/889121 Fax : 02381/880084 Fido : NO FIDO ANYMORE Data : 02381/880081 Traily: 02381/880082 Login : box | uucp | nuucp