Date: Mon, 19 Dec 1994 10:19:34 -0600 (CST) From: "Traci J. Ingram"
<> Subject: Of Line noise, The Phone
Company, and Your Modem

Here is a fairly recent document from Supra, regarding telephone line
noise (tel_line.txt, revision 1.6), in which there is given blanket
permission to re-post (if this already has been uploaded, please
disregard; however, I could not find it in comm/info).  Supra Sysop
Patrick Moore says:

*** This document will be updated at random points in time, so please
check the Supra BBS {503-967-2444} to assure the most current

This document is a combination of data from engineers at both Supra &
AT&T, electronic mail from other people involved with the modem &
telephone industries, and my own knowledge and education relating to
data communication and transmission lines.  Please feel free to post
this document in its original and unedited form.

Patrick Moore Supra Technical Support - Senior Staff System Operator of
Supra BBS Original Document: Feb. 25, 1993 Latest Update:     Aug. 5,

****************** Reposted to Info-Mac by Traci Ingram     | or |     101-8673@MCIMail.Com

***  Of Line noise, The Phone Company, and Your Modem  ***

TEL_LINE.TXT  rev. 1.6  (PHM 8/5/94)  from Supra BBS (503-967-2444)

* NOTICE to users on PBX, "Lease-Line", and other non-PSTN phone
connections *

Your modem is designed to function only on standard phone service (the
common analog 2-wire Hybrid single subscriber line).

PBX or similar use on a phone jack connected to a digital phone port
(as with many AT&T, Northern Telecom, and similar multi-line office
phone systems) you will need to do one of the following things to use
your modem:

1. Have your phone system administrator install an "analog port" card
into the PBX to allow standard telephone devices (like your modem) to
be connected. 2. Talk with your phone system administrator about
getting a digital to analog jack converter box, or a digital telephone
with an analog modem jack, to allow your modem to be connected via the
existing type of telephone service. 3. Have a separate analog 2-wire
hybrid phone line installed by the local phone company for use by your

"Leased-Line" (Common types)

2-Wire Hybrid:  Not truly supported, but can be forced to connect by
simultaneously issuing ATA on one modem and ATXD on the opposite modem.
 (This method is also usable with cases of "simulated" phone lines. 
Like connecting a terminal or other device via two modems connected
with twisted pair wiring.) 4-Wire:         Not supported via any

Other phone system using non-standard telephone connections should be
avoided unless it has been demonstrated that common 2-wire analog
telephone devices are correctly supported.

"Line noise" is commonly used as a general term to describe all types
of signal loss and signal damage that occur in the phone system,
however, this is not the correct terminology to use in a discussion of
phone system problems. According to the Standards set by the
Telecommunications Industry Association, "line noise" refers to a
single specific impairment, properly labeled "1004Hz Signal to 3KHz
Noise".  (cf: EIA/TIA-496-A)

To the layman, "line noise" also refers to the white or pink thermal or
diode noise injected into a channel by an amplifier (or amplifiers)
with limited negative feedback.  With this in mind, a more correct
phrases would be "telephone line defects", "transmission media
problems", or  "unfaithful signal transfer".  This broader scope of
problems related to correctly passing the original signal from one
modem to another via the telephone network is the major topic of this

In reading the following information regarding your subscriber line,
you must remember that you can only address the type of phone service
you pay for and the long-distance carrier your call is routed through. 
You normally have NO control over conditions that may be present in
local office switches or calls placed in your "LOCAL LONG DISTANCE"
area.  This means that regard- less of how much the phone company may
do to your line, there may still be places you can't get a good
connection to.  Also, if the whole problem turns out to be the local
switching equipment, you are just plain out of luck (unless you can get
the local phone company to replace equipment that runs between tens of
thousands to hundreds of thousands of dollars).

POTS (Plain Old Telephone Service)

This is the type of phone service most people have.

This is what it you get:  (Please note that the legal requirements for
voice lines aren't even this good, but this is an average example for
telephone service in the USA.)

-most 300Hz to 3000Hz sounds are passed.  (Although the system is not
required to fully pass this range.) -You aren't likely to get any more
than -10db loss from the local telephone office to your phone jack.
(Measured from a 1mW 1kHz tone source.) -VOICE use (you should be able
to hear another person's voice using normal, FCC approved, telephone
devices via this service).  (No implied ability to use Modems, Faxes,
or other "non-voice" devices over this service.) -Pulse Dialing

Many URBAN users serviced by newer switching equipment MAY also get
some or all of the following:*

-TouchTone Dialing. -Certification for 2400bps data carrier use.
-Certification for 4800bps data carrier use. -Certification for 9600bps
data carrier use. -Certification for 9600bps fax carrier use.

A small (but growing) number of URBAN users may also have:*

-"3002" Class lines able to handle data carrier rates up to 24000bps
-Special lines able to handle carrier signalling rates above 24000bps.

*Remember that just because your LOCAL telephone company may support
these modes, it DOES NOT mean your long distance or LOCAL long distance
service will.  (It is NOT uncommon, especially with VFC rates, to find
that one LONG DISTANCE provider may be able to provide higher rate
support than another.  Also, you find differences in rate support to a
given area varies from one provider to the next.)


These types of service are offered as an "upgrade" from POTS (often at
a much higher installation fee, and an additional monthly service fee).
The services supported and/or promised by such upgrades will vary
widely from one telephone service provider to the next and are not
supported at all in some cases.  In general, they will provide a
"cleaner" telephone connection that is able to support fax and data
communication to a specific carrier rate, and normally have no more
than a 5db loss.  These services may also provide a specially installed
line with no LOADS and no BRIDGE-TAPS. Ideally, these lines will be
able to supply a 600 ohm balanced line, and should respond very well
when devices (like FCC approved Telephones, Modems, Faxes, etc.) are

Items and Conditions that commonly degrade telephone lines:


These are "extra" wires connected to the wire pair coming to your home
from the local telephone office.  Often these are the result of normal
service and repair procedures that occur over the years.  These don't
normally create a problem for VOICE use (although they have been traced
to things like radio stations or buzzing noises in some cases). 
Devices using complex signaling tones and patterns, however, can be
greatly affected by these "extra" signals on the phone line.  Theses
extra wires act like antennas and pick up noises from a wide range of

LOADS (Step-Up Transformers)

These are commonly used to increase the volume of voices on longer
phone lines.  They cause distortion of the rated 300hz to 3000hz band
pass, and can cause the frequencies to be shifted up or down from their
original values.  The volume of these frequencies is also not increased
equally. In some cases, the line can become poor enough that TOUCH-TONE
dialing may not be supported, and only PULSE dialing is possible. 
LOADS are most commonly used in older, rural settings, but still do
exist in some urban installations.  LOADS also serve as a means of
introducing high levels of noise onto the line, both by increasing the
level of noise on the line along with the voice level and by picking up
environmental noise from the area directly around the transformer
itself (other transformers, power supplies, high voltage power lines,


This is the composite of all of the types of loss on your phone line
which include the following:

-Pure Loss

This natural loss in the phone line caused by the resistance of the
wire and connections between the local office and your home.

-Return Loss

This is a measurement of the signal loss on the line.  This is affected
by the line itself (see above) and the devices you have connected to
it. It is basically a measurement of the impedance match between the
local office and the line and the devices you have connected (problems
in any part will affect the over-all measurement).

-Frequency Distortion and Attenuation.

This is basically the loss of volume and shift in frequency on the
line. Problems often arise in high-speed data transfers because only
part (or in very poor conditions none) of the original signal gets
transferred faithfully to the receiving end.  In some cases many
frequencies have been "reflected" back on to each other, further
damaging the signal.


-Quanitazation noise

Digital links have quanitazation noise which is inherent in A/D

-Thermal Noise

This is naturally occurring noise sourced from normal electrical decay
and energy transformation.  Local loops, if properly engineered, have a
thermal noise of 22 dBrn.  But it is not uncommon to find it at higher
levels. (22 dBrn in "reference noise" is equal to -68 dBm.)

-Digital Compression & Multiplexing

Compressing or "slicing" the signal to get a higher number of voice
signals on one carrier signal or channel.  (This is just fine for
voice, as the human ear does not notice the small gaps in the signal
this creates, but modems can't deal well with gaps in the data

-Power line hum

Low frequency noise conducted into the phone line as it passes power
lines. (Also it is possible for power hum to be passed into the phone
line from power sources in the phone system itself.)

-Phase Jitter and Phase Shift

The shifting or bouncing of the carrier signal in relation to time.


Noise or sounds from other lines in the phone system.  (In very bad
conditions this can be loud enough you can hear other lines ringing and
even hear other voice or data signals.)

-RFI (Radio Frequency Interference)

High frequency electrical signals conducted into the phone lines.  (In
very bad cases radio stations or other radio transmissions can be
heard.)  If you can figure out the source of the signal (like the name
of the radio station), then you should contact them in regard to curing
this problem.  In the US, FCC regulations require the person in charge
on the transmission to fix this problem at his expense. (In the case of
a radio station, contact the station engineer.  He can normally
construct a filter to remove the signal from your phone line.)


Bandwidth is the frequency range available on a given connection.  At
low rates (4800 and lower), this is normally not a problem. 
Connections in the the 9600 to 21.6k carrier rate range something
expect near a 300Hz to 3KHz range to be available. (This is about the
range ideally available with most telephone service.)  Higher VFC/V34
rates (24k to 28.8k) require nearly the full 0Hz to 4KHz range to be
available to operate successfully.

Common Tests to ask to have done by the phone company:

-Frequency Test (also called Frequency "Slope" or Frequency "Twist"

This test shows both total loss in signal, and the amount of difference
in signal loss at different frequencies.  A 1mW 1Khz signal is measured
for total  loss (in dBm) and then compared to readings taken from
signals at 400hz and 2800Hz.  For a VOICE line, the high and low
frequency values should be within a +3dBm to -5dBm range of the value
of the center frequency (a DATA grade line should be closer to the
reading of the center value). In conditions where this problem is quite
bad, you will hear the voices as being "tinny" and it may be more
difficult to recognize voices over those lines.

-Loss and/or Return Loss Test

This will test the loss of a signal transmitted from one end of the
line to the other.  (In the case of a Return Loss Test, it tests the
loss of a signal transmitted into the line and reflected back from the
other end. -45dBm is a reasonable level, -60dBm is not acceptable.

Common Problems, Tests, and Repairs for end users to address:

-Try the modem/fax with ALL other devices removed from the line.

This tests for two common problems:

1. NOISE -from devices like cordless phone and "neon-display" phones
most commonly, but also can be sourced by any telephone line device.

2. Signal Loss -from devices that "pull" power off the phone line. 
Phones and devices that take power from the line when not actively in
use for dialing memory, status lights, and lighted displays (although
most lighted displays are only taking power when the handset is picked
up) will lower the power level on the line, and will also lower the
signal levels being passed on the line.

-Check the condition of the phone wiring in your home.

"High Twist" (the wire pair or group of pair is twisted accross the
other wires several times per foot) or "Data Grade" phone cable, rather
than standard "low twist" or non-twisted phone cable, wire is better
for high-speed data transfer.

"Silk" cable (that flat cable that runs between the phone device and
the phone jack) should be kept short (under 8 feet).  It is better to
put in a new phone jack (preferably wired with "High Twist" cable),
then to connect the modem via a long "silk" type extension cable.

Wiring should be solid or stranded copper (avoid copper-plated steel or
aluminum wire).

Wiring runs should advoid passing near lightning (esp. flourescent),
motors, and other electrical devices as much as possible.

Wires should be at least 26 gauge, 22 gauge is better.

Wiring connections and terminal blocks should be clean (light polishing
with a brass wire brush or fine sandpaper can work wonders), dry, and
free of oxidation.

Insulation should be in good condition.

Try a noise filter (here are some suggestions):

1. Non-Intrusive Inductive Filter:

This is very simple to install, and has the lowest signal loss (none to
bother measuring for the desirable freqencies). To make such a filter
simply loop the telephone wire multiple times (three or four is enough)
through an iron or ferite core.  (Such cores are commonly available
from electronic and computer supply shops and can easily be scrapped
from old equipment as well.  Those "bulges" on many computer monitor
and keyboard cables are ferite cores.) (You may also find it helpful to
over-all operation to put similar filters on the modem's power cord and
serial cable.)

2. Intrusive Inductive Filter:

This filter works by the same general concept as the first filter
listed, but requires the telephone line to be cut and then this filter
would be inserted at the splice.  Such filters also add resistance to
the line, and may affect the impedance match of the modem to the
telephone network. These are a "second best" choice.

3. Intrusive Capacitive Filter:

This filter works by placing one or more capacitors across the phone
line or a network of resistors and capacitors on the line.  These
filters are generally a poor choice for use above 2400 baud (which is
NOT to say they don't work well in specific cases), and they normally
reduce signal strength by ten to fifty percent.  In general, I would
not suggest using such a filter until all other options have been

Wiring runs should be kept out of direct sunlight and weathering.  (If
wiring must be done outside, use phone wire rated for outdoor use.)

Remove wiring runs no longer in use (this does not mean to rip out any
phone outlet you aren't using, but it is a good idea to remove wiring
that has already been disabled or covered up by remodeling -many older
homes and rentals have phone wiring that go nowhere!)

*WARNING* If you find equipment that you don't understand connected to
the phone lines in your home, or "extra" wiring to transformers or
grounding rods/water lines, do not disturb it without talking to the
phone company about it first.

To understand V.FastClass (V.FC) & V.34 (aka V.Fast) connectivity a few
items must be clarified:

-  Repeatable results with V.FC/V.34 modems on real-world phone lines
are sometimes difficult to obtain.  Unlike V32.bis modems, V.FC/V.34
modems will actually modify their data coding (symbol rate) and
transmit levels during the connect phase.  Very minor differences in
the phone line can lead to quite different results.

-  A 28.8K connection on some real-world lines may not occur very
often, if it occurs at all.  Preliminary testing in the lab indicates
that "normal" phone line attenuation makes a 28.8K connection difficult
even in the absence of line noise.  This is not surprising considering
that the V.FC & V.34 coding schemes are relying on the bandwidth of the
phone system being greater than what is actually available in some
areas.  Very high-speed connections assume that there is a substantial
amount of digital technology in the system (thereby normally increasing
the usable bandwidth).  Some users are going to find that their local
phone system effectively limits them to 16.8K or 19.2K.

-  Both the originating and the answering modem are adjusting to their
line conditions during the connect phase.  The fact that a connection
to a given modem at a given location occurs at a lower rate (19.2k for
example) may have nothing to do with the modem the call is placed with,
or the location it is placed from, as the modem on the other end may be
adjusting the baud rate down.

*** This document will be updated at random points in time, so please
check the Supra BBS {503-967-2444} to assure the most current

This document is a combination of data from engineers at both Supra &
AT&T, electronic mail from other people involved with the modem &
telephone industries, and my own knowledge and education relating to
data communication and transmission lines.  Please feel free to post
this document in its original and unedited form.

Patrick Moore Supra Technical Support - Senior Staff System Operator of
Supra BBS Original Document: Feb. 25, 1993 Latest Update:     Aug. 5,


PBX            Private Branch eXchange - the type of phone system used
in most offices that supports a private network of tele- phone devices,
and may or may not connect to the PSTN in some manner. PSTN          
Public Switched Telephone Network - The common consumer telephone
system made up of local and long distance carriers, where each
subscriber has a unique and directly accessable number. Leased-Line   
A private telephone line (in some cases, just a wire pair connected
between two points) leased from the phone company that is used for
voice, data, or control signalling.  This service is available in many
forms, and availability varies. Most common consumer modems (including
the SupraFAXModem serries) are not design to operate on this type of

Information below is pasted in from the Internet and is as-is, but may
be useful.

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> START of Paste

>                           NOISY CONNECTIONS? >          (Minimum Line
Testing Requirements as set by the FCC.) >                  Lloyd Kautz >     From "The CPC Newsletter" A monthly publication
of the Connecticut >     IBM Personal Computer Users Group. > >        
                          * > >     Your local telephone company is
required by the FCC to provide a minimum >level of quality on the lines
it maintains for your use. > >    Many lines do not meet these
standards, and so are a source of considerable >noise in transfer of
data between computers. A noisy line can make it very >difficult to
upload messages and files to a BBS. > >     If a SYSOP complains to the
telephone company about line noise, there is >usually only one option
offered: line conditioning. This is an extra cost item >that they would
like to sell you. However, if their line to your computer meets >the
minimum standards, there is no reason you cannot have error-free
transfers >at 2400 bps and, if an error-correcting protocol is used,
reliable operation at >9600 bps is the norm. > >     If the telephone
company seems reluctant to test your line and assure that >it meets the
minimum requirements (and they will), request that they run the
>following tests and provide you with the results. Let them understand
that if >they do not comply, you will refer the matter to the FCC. They
will not be >happy. > >     The following tests should be run: > >    
1. Frequency sweep; 300 to 3000 hertz. >     2. Roll off 5 to 2500 hz
with -2 to +8 range. 300 to 3000 hz: -3 to +12 >        range. 1000 hz
tone loss  -16, 1% either direction. >     3. Signal to noise: 24db. > 
   4. White noise C, message Dbrnco. >     5. Envelope delay
measurement. >     6. Phase jitter, not to exceed 10%. >     7. Impulse
noise test. Min. 15 minute test. > >     Don't worry if you do not
understand what all these tests mean, the >telephone company knows.
Their standards are all available from the FCC upon >request. When the
lines provided to you meet these minimums, you will >experience
reliable, accurate, and noise free data.

****** Comment on the Data Above ******


I'm not sure what the exact relationship between the FCC requirements
and actual Local Exchange Carrier (or LEC, the telco) tariffs are these
days...  but it really doesn't make much difference for our purposes. 
The telco has a tariff filed with whatever agency regulates it, and
that tariff is the final word on what service they are able to sell. 
It is in fact going to look almost identical to the FCC's recommended
requirements, which means almost all of them are going to be the same
and we can discuss them as if they are the same.  However in any
particular case you might want to get a copy of the tariff from the
telco (they are required to make it available).  It makes a great
little red flag to wave in the face of the bull.

With that said, if you have a problem with a telco that won't fix your
line, keep in mind that the telco has very little experience in what
you are asking them to do.  (It isn't that they are total dummies, even
if it sounds like it to you.  You probably don't know as much about it
as they do, either, and that doesn't make _you_ a dummy...) So you may
well have to encourage them a little. The best way to do that is to
continually remind them that your modem was designed to operate over a
standard voice grade dialup line.  If it is not working then either the
modem is bad or the line is bad.  You have verified that the modem is
good, so is apparently a bad line.

It happens that a v.32 or later modem is a _very_ good test of a
telephone line, and the telco may not have test facilities that can do
as good a job!

>       If a sysop complains to the telephone company about line noise,
>there is usually only one option offered; line conditioning. This is
an >extra cost item that they would like to sell you. However, if their
line >to your computer meets the minimum standards, THERE IS NO REASON
NORM. (Emphasis >added.)

To update that...  operation at 14.4 Kbps is the norm.  With a 28.8
modem do not expect to get 28.8 (be glad if you do), but it should be
better than 14.4 on most of them.

>       If the telephone company seems reluctant to test your line and
>assure that it meets the minimum requirements (and they will), request
>that they run the following tests and provide you with the results.
Let >them understand that if they do not comply, you will refer the
matter to >the FCC. They will NOT be happy! (Emphasis from original

I would change that a little.  Work with them before getting to the
point of threats.  When it does get to that, threaten to go to their
state or local regulatory agency, not the FCC.  That will almost always
have much greater effect.

>       The following tests should be run:

Here are a few technical comments on these tests...

>       1)      Frequency sweep. 300 to 3000 Hz.

You want to see what it looks like over 200 Hz, not just a three tone
run (usually done at 404Hz, 1004Hz, and 2804Hz).  Don't be upset if the
actual frequencies are different by a few Hertz.

>       2)      Roll off. 500 - 2500 Hz, with -2 to +8 range, >        
      and request 300 - 3000 Hz, -3 to +12 range. >               1000
Hz tone loss: -16, 1% in either direction.

This is confused and from two different sources.  On the first two
lines the range is given in "loss," which is negative gain.  What it
means is the level from 500 to 2500 Hz can have as much as 8 dB more
loss than the loss at 1000Hz.  The level can only be as much as 2 dB
higher.  The figures given for 300-3000Hz are the same (12 dB more
loss, 3 dB more gain).

The third line where it says -16 means just that.  It can be 16 dB of
loss from one end to the other, and the two directions should be the

>       3)      Signal to noise ratio. 24 Db level.

Higher numbers are better here.  The best you could possibly get is
about 34 dB, and commonly 28-32 dB is seen.  (It is interesting to note
that of all these tests, the only one with different specifications for
the "data conditioned" line that is often offered, is this one: 
increased to 28 dB.  Not included here, but also different are the
specs for harmonic distortion.  They are not important, but the signal
to noise ratio is.)

>       4)      White noise C, message Dbrnco

That isn't white noise.  It is C-message weighted noise, dBrnC0. A
lower number is better.  I can't remember the exact specs for a local
loop, but I'm sure that 30 dBrnC0 would be high, and lower numbers are
better.  (Long distance calls will be higher, and a satellite trunk
will be as high as 54 dBrnC0.  So don't worry much about it...)

>       5)      Envelope delay measurement.

The telco probably can't do this test, and it makes no difference in
the operation of a dialup modem.  There is no specification for
envelope delay on a Plain Old Telphone Service (POTS) line.

>       6)      Phase jitter. Not to exceed 10%.

I would get upset if phase jitter was greater than 4-5 percent on a
local loop, but it would be in specs.  10% is the end to end spec, but
no part of the circuit can be that high.  It is very unlikely you will
ever see a problem with this parameter in a digital telephone system. 
(Note that if your local loop is all copper cable, this test is a joke
because a cable can't cause phase jitter.)

>       7)      Impulse noise test. Minimum 15 minute count.

This one is important, and there is a lot more to it that just an
impulse noise count.  Normally a test set will do impulse noise, phase
hits, and gain/dropout hits all at once.  The spec for each is 1 hit
per minute over a 15 minute period.  But that is absolutely terrible
performance!  (The spec was written in the 1930's.) The important thing
is the threshold for each test:  72 dBrnC0 for impulse noise, 2,3, or 4
degrees (the smallest available on the test set) for phase hits, and
2,3, or 4 dB (again, the smallest available) for gain hits.

Probably each and every event counted (you may get one of each kind in
a single event) is going to show up on a modem as an error or group of
them.  So what you want to see with this test is no counts at all over
15 minutes.

>       Don't worry if you do not understand what these tests mean, the
>telephone company knows.

But you may not be talking to the one or two people they have who
knows.  If the person you talk to does not understand, ask for someone
who does.  Go up the management chain if need be to find someone with
the authority to give your problem to a tech who does know this stuff.

> >       I called the phone company. The technician claimed never to
have >heard of these tests. She performed something called an MLT test
and >pronounced the line to be OK, telling me that the problem was with
my >modem. I spent 3 hours on Gopher trying to find information on
these line >requirements with no luck.

I'm scratching my head and can't figure out what an "MLT" test would
be, unless he means "modem line test"...

However...  it might be wise to check out the most likely cause of your
problem, which it happens that you can test for much easier than the
telco can!  Make a connection at 2400 bps with no error correction and
no compression, and see if you get garble that has repeating patterns
of the same characters in it.  Here is a real example:

Local> {{{r{{{{m{xD{{rw3{{r{{{m{t({{xD{{{{v{{{t(t^O5rw3{{v:{{ Xyplex
-701- Command syntax error

Note that "{{{", which is the most common pattern seen, repeats a lot. 
Note the two times that "w3" shows up...  that is another common
pattern.  The real point though is that it isn't random. The same
pattern keeps getting in there.  Even if it isn't very often and there
is a lot of other garbage, the repeating pattern is the important part.
 The reason is that normal noise is random, so what is causing that
problem is electrically not random.  It is a telco equipment problem,
and it is something that repeats itself in just about exactly the same
way each time.

If you see repeating patterns, tell them you are experiencing trouble
because their equipment is suffering "controlled clock slips" at a
digital interface.  Whether or not that makes sense to the person you
tell it to, it ought to set them back a couple paces that you said it!

Especially if you see the "{{{" patterns...  keep after them and don't

Please feel free to email me with questions or more info on what you
have or what the telco says.  The difference between me and the telco
tech you work with is that I'm an IXC (long distance telco) tech, and
we do this kind of thing all the time.

Floyd - --        A guest on the Institute of
Marine Science computer

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> END of Paste