Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cwjcc!neoucom!wtm
From: wtm@neoucom.UUCP (Bill Mayhew)
Newsgroups: comp.dcom.modems
Subject: Re: modem line noise
Summary: type of modulation
Keywords: modem line noise
Message-ID: <1712@neoucom.UUCP>
Date: 6 Sep 89 15:47:58 GMT
References: <89Sep1.235928edt.19631@me.utoronto.ca>
Distribution: na
Organization: Northeastern Ohio Universities College of Medicine
Lines: 55


Why are 2400 bps connections noisier than 1200 bps on the same
line?

Answer:  both 1200 and 2400 transmission rates use the same 300 bps
carrier but differ in the number of phase-angle modulation states
that the carrier can have.  For 1200, there are 4 states (300 * 4 =
1200), and for 2400 there are 8 phase states (300 * 8 = 2400).  The

If you draw an S-plane map, the configuration is something like:

          ^                    ^
          |                    |
          |                *   |   *
       *  | *                * | *
          |                    |
      ----+---->           ----+---->
          |                    |
       *  | *                * | *
          |                *   |   *
          |                    |

Some of the fancier modems such as the AT&T 9600 baud V.32 modem
have so-called constellation generators that can be attached to an
X-Y mode oscilloscope to generate patterns similar to the above.
What you see is actually a jumbly ring in the vicinity of the ideal
locations as the bits are recieved.  Trouble happens when the
jumbly rings overlap; this indicates that there is phase jitter
somewhere in the communications network.  Jitter can come from
poorly equalized lines that have group delay (some frequencies
propate faster than others), from digital links that insert
administrative packes at irregular intervals, or from a defective
defective modem component.

The reason that 1200 works wehn 2400 doesn't is that the distance
between the centers of the phase states is greater for 1200 than
2400, thus it is easier for the demodulator to make a decision
to which phase state a particular buad belongs.  I have seen a
number of instances where a noise free line at 1200 bps is
completely unusable at 2400.  Note that in such a case, MNP error
correction is of no help, as MNP is really only effective is the
noise is sporadic.

The axiom of, 'you get what you pay for', applies at least
partially to modems.  More expensive units often employ more
adpative equalizers and tunable filters that can help compensate
for group delay and echoes.  Often the special features are
integrated into a chip and may not be readily apparent upon visual
inspection.  Often the glossy sales brochures don't provide much
real information about a modem's signal processing capabilities.
Field testing seems to be the best method for guaging modems'
performance.


Bill