Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!csd4.csd.uwm.edu!uakari.primate.wisc.edu!ames!apple!well!elc
From: elc@well.UUCP (Eric L. Cook)
Newsgroups: comp.dcom.modems
Subject: Re: severe line noise
Summary: Comments and ** WARNING ***
Keywords: modem noise filter
Message-ID: <13429@well.UUCP>
Date: 31 Aug 89 21:18:35 GMT
References: <434@tron.UUCP> <9411@chinet.chi.il.us>
Reply-To: elc@well.UUCP (Eric L. Cook)
Organization: Whole Earth 'Lectronic Link, Sausalito, CA
Lines: 46

On the Chinet posting about a "Modem Noise Killer" I have a few
comments about what I saw. First let me say I design modems for a
living.

What you are doing is not filtering noise but changing the
impedance match on the phone line. A nominal phone line should
be about 600 ohms, but a lot of phone lines run in the 750 to
1200 ohm range. As you can see there is not a very tight control
over this spec. Impedance matching is important to a modem as
the modem does a "4 wire to 2 wire conversion". That is 4 wire,
(1)TX, (2)ground and (3)RX, (4)ground to 2 wires, Tip and Ring.
In a perfect conversion when you transmit on the TX line you will
not see your transmit signal on your incoming receive line. 
( Its like if you where talking you could not hear yourself talk
but you could hear the other person talk. ) Any time there is not
a perfect match you can start to hear yourself talk. To a modem
this shows up as noise on the receive signal affecting the signal
to noise ratio.

A 300 BPS modem needs a minimum 8 DB's of signal to noise ratio
for a bit error rate of 1 x 10^-5. With a 2400 BPS reception you
need a minimum of 17 DB's of S/N ratio for the same error rate!
So you can see that noise can play a important part in modem
transmission.

The reason you are not filtering noise for better reception is
that you are filtering the receive signal as well, and in doing
this you are not improving the S/N ratio. 

!!!!!!!!!!!************   WARNING  **************!!!!!!!!!!!!!

(1)  These changes that you are making ARE illegal. You are
affecting a "public" network and the FCC has very heavy fines for
such violations.

(2)  I see people quoting 100VDC, or 200VDC caps. The ring signal
is 100 volts ***RMS***, not DC. A minimum of 250 volts is needed.
Also the cap must be non-polarized.

(3) If you live where there is lightning you are exposing
yourself to a shock hazard by exposed metal parts. ( Resistor
bodies, wires. )

There is the information, you can make the decision for yourself.

Eric Cook