berger@clio.las.uiuc.edu (03/12/88)
It sounds like your Zenith, Everex, and Amazing modems have the
problem, not the NEC modems. I've never seen this problem with
NEC modems used to dial in. It may be an artifact of your
phone switching system, too.
Mike Berger
Department of Statistics
Science, Technology, and Society
University of Illinois
berger@clio.las.uiuc.edu
{ihnp4 | convex | pur-ee}!uiucuxc!clio!bergerdricej@drilex.UUCP (Craig Jackson) (03/20/88)
In article <2431@s.cc.purdue.edu> add@s.cc.purdue.edu (Lon Ahlen) writes: >We have experienced a problem with some manufacturer's 2400 bps modems >connecting to our bank of NEC N2420/30-P modems. > >When one of these modems dial into our NECs from an on-campus phone, our >modem answers and presents carrier before the originating modem has >sensed ring-back. The originating modem will not sense the carrier from >our NEC and the connection fails. > >This problem has been reported by users of "Zenith", "Everex" and "Amazing >Modem". I've had this very problem with true-blue Hayes 2400 modems. After some complaining, I finally got a chance to speak with one of the developers. It seems that the Rockwell chip set that many 2400 baud modems use cannot listen for call progress and carrier detect at the same time. So what modem manufacturers do is listen for a ring-back in the call-progress band, and then switch the receiver to listen for carrier detect. Unfortunately, this collides with modern phone systems. On most telephone exchanges, the 'ring-back' has nothing to do with the actual ring current. However, on older exchanges you can depend on them to come roughly at the same time. They're both timed using a motor-driven interrupter. On electronic exchanges, there's really no hard reason for them to come at the same time. So frequently they don't. On our Dimension (TM) system, the ring-back appears to come up to three seconds after the actual ring. Certainly, a modem has time to answer in three seconds. If you turn off call-progress detection (at least on a Hayes) everything works. The moral is: many modern modems have call-progress detection, but it is flawed, and therefore unusable. Of course, the modem manufacturers are not quick to warn anybody, and therefore there's a lot of software out there that uses the call-progress-detection. BTW, I was also told that the Hayes V-series modems use a different chip set which do not suffer from this problem. I have no way of verifying this. -- Craig Jackson UUCP: {harvard!axiom,linus!axiom,ll-xn}!drilex!dricej BIX: cjackson
berger@clio.las.uiuc.edu (03/22/88)
I suspect this is a problem with the phone switching system, and not with
the modem. The timing for answerback on the NEC 2420/30 follows the
Western Electric specifications precisely - and NEC doesn't use a
Rockwell chip set in the modem. Clearly, it would be next to
impossible to expect a modem to work with every phone switching system
in the world, if the switching systems don't adhere to standards.
Mike Berger
Department of Statistics
Science, Technology, and Society
University of Illinois
berger@clio.las.uiuc.edu
{ihnp4 | convex | pur-ee}!uiucuxc!clio!berger