Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!purdue!ames!lll-winken!scooter!neoucom!wtm
From: wtm@neoucom.UUCP (Bill Mayhew)
Newsgroups: comp.dcom.modems
Subject: Re: 2400 baud modem recommendations?
Summary: Look for a modem with an adaptive equalizer
Message-ID: <1620@neoucom.UUCP>
Date: 9 May 89 14:31:13 GMT
References: <1052@calvin.EE.CORNELL.EDU> <18032@cup.portal.com> <9020@polya.Stanford.EDU>
Distribution: usa
Organization: Northeastern Ohio Universities College of Medicine
Lines: 69


Not all modems are created equal in terms of noise rejection.
There are noticable differences in the quality of filters that are
used to recover the received carrier.  The second harmonic overtone
of the transmitted carrier can be problematic where there is a
line impedance mismatch that cuases the originate carrier to be
reflected back into the receiver of the originating modem.

There are three factors that affect performance.  First is the
quality of the digital or analog filter that is used to reject the
overtones of the originate carrier.  Second is the use (if any) of
an adaptive equalizer to assure flat frequency response over the
passband of the received carrier.  Unfortuantely, advertising
literature hardly ever states performance figures or design of the
filtration and/or equalization used in the receiver.  I don't think
I've ever seen second harmonic rejection specified.  The third
performance parameter is echo cancellation; this shouldn't matter
on a v.22.bis modem as decent filtration in the receiver should be
enough, as the xmit and receive carriers use different fequency
bands.

The line between work and my house is very problematic.  Here are
some modems that I have found that work well.  Your results may
vary.  In all cases, the modem at the other end was a Trailblazer;
I used a variety of modems on my end.  The line is so bad that a
Hayes Samrtmodem 1200 will not even connect on the line.
Here are the best:

1.  IBM PS/2 internal 1200 baud modem.  This modem does not have
    MNP correction, but essentially completely corrects for line
    impairments (apparently by using adaptive equaliztion).  The
    unit is made by Racal-vadic for IBM.  I have not been very
    impressed with other Racal-vadic stand-alone modems.

2.  US Robots Courier HST.  Not tested in HST mode, as I didn't
    have two of them.  MNP mode works well.  Corrects almost all
    line problems even without MNP mode engaged.  Tested both 1200
    and 2400 baud.  Trailblazer would only support up to MNP level
    3 on with the HST, due to limitations of the trailblazer.

3.  Trailblazer Plus.  Noise free in all modes.  Good response in
    MNP mode.  As both ends employed trailblazers, the results with
    PEP were very impressive.  Interactive turnaround in PEP mode
    with short packets is good (both ends using 4.0 firmware).
    uucp transfer stats average 1200 char/sec for ~100K files in
    PEP mode.  I still prefer to use 2400 baud for interactive
    editing in vi.  I use PEP mode to read news.

4.  AT&T 2224 CEO.  Resonably decent performance with PEP enabled.
    Has annoying ~1.5S delay as modem switches from short to long
    packets after ~32 characters are received at maximum baud rate.
    Will occasionally refuse to connect at 2400 baud over poor
    circuit.


After trying out a bunch of modems, I decided on the Trailblazer as
it is the most flexible and seems to have the best noise rejection
at 1200 and 2400 baud even without the MNP mode engaged.  I use the
trailbalzer a lot for uucp transfers, and have cut my connect time
from about an hour to about 5 minutes a day.

If you are looking for v.32 compatibility, the new trailblazer 2500
or the UDS v.32 modem that includes v.22.bis are likely to be good
performers.  I tested the older UDS v.32-only modem and found it to
be the best performer, though it was by far not the most expensive
v.32 modem.  I have not tested the trailblazer 2500.

Bill
wtm@impulse.UUCP