Index Home About Blog
From: (Floyd Davidson)
Newsgroups: comp.dcom.modems
Subject: Re: Line Quality isn't the enemy...line upgrades are...or, tilting at 
	the Ma Bell Windmill.
Date: 14 Sep 1996 00:10:52 GMT

[emailed and posted] wrote:
>Floyd Davidson wrote:
>>If you consistently get 28.8 connections to _anyone_, then your
>>line must be virtually perfect
>No. That ISP has Couriers.  And have probably gone around with the
>telco about line quality, until they got some (line quality that is).

But if you get 28.8 connections consistently to anyone, then
your line definitely far exceeds the minimum parameters the
telephone company has guaranteed you will get when you buy 
their service.

>> I wouldn't spend much time bothering them about quality...  they
>> might give you line to someone who appreciates it!  :-)
>I wasn't complaining to the telco.  I was inquiring about line quality.
>I was amazed at their response.  The problem is with one ISP who falls
>back on "it's the phone company I can't do anything about it" when I
>complain to them about connect speed.  They contracted for a 28.8 line.
>I want a 28.8 line.  I don't care that they have problems to solve in
>order to comply with the contract.

But Paul you have not contracted for a 28.8 line, and neither has
the ISP.  The telco guy is about as non-technical as you are, but
in a different direction.  He didn't understand your question and
you didn't understand his answer.

The technical answer is that the telco sells you a line which must
meet the minimum specifications guaranteed by their filed tariff.
That tariff, for example, says that you will _probably_ get a line
with a given signal level, a given bandwidth, and a given S/N.
But it gives minimums for those parameters too.  2400Hz bandwidth
(400 to 2800 Hz), a signal level that is 8 dB lower than "normal",
and an S/N of 24 dB (average is probably 30-32 dB).  Plus, that is
_your_ line, and says nothing about what happens when you connect
to another line, which might be average, or better or worse.

When the telco told you they don't guarantee any modem speed, they
meant it.  How well a modem works over a minimally spec'd line
is not the telephone companies problem and they have no control
over it.

Now, a v.32bis modem is designed to connect at 14.4Kbps over a
minimally specified line.  If you cannot get a 14.4Kbps connection
using a v.32bis or v.34 modem, then either the modem is bad or the
line is bad.

But if you can get a 14.4Kbps connection, then your telephone line
is at least as good as the phone company said it would be.

Of course your phone line might be one heck of a lot better than 
that too, and modem companies have provided us with v.34 modems
to take advantage of that.  If you get 28.8Kbps connections, then
call your modem company up and thank them!  Call the phone company
and thank them too.  But if you don't get anything faster than 
say about 16.8Kbps, just remember that nobody said you should!
And nobody took your money by saying you will.

Floyd L. Davidson          Salcha, Alaska

Newsgroups: comp.dcom.modems
From: (Floyd Davidson)
Subject: Re: testing phone line quality (techie people read please)
Date: Sun, 22 Jun 1997 06:43:50 GMT

[emailed and posted]
In article <01bc7d4b$4d9411e0$c1ad61cc@guardian>,
Guardian <> wrote:
>I'm trying to determine the following
>a) quality of phone line entering house
>b) degradation when using house wiring
>I used a radioshack digital multimeter, and tested (a) at 56.6 V DC, and

Multimeters are great devices for a number of things, but they don't tell
you anything useful about a telephone line.

>What should I be testing, and what should I look for? If this device
>doesn't have the appropriate setting for testing phone lines, what should I
>look for?

Well, a few years back I'd have said that for about $12,000 you
could get a half way decent test set, and for about $20,000 you
could actually get a decent one.  But today, go buy a 28.8K or
33.6K v.34 modem and it will tell you more than virtually
anything else.

Try the modem several times with calls to different distant
modems.  Do this first from as close to the entrance point
(demarc) as you can get, and disconnect *everything* else.  Do
not leave any wiring connected other than the line to the modem.
For the second stage, you can hook all of the house wiring back
up and take the modem to whatever location you wish to use it
from on a normal basis.  Repeat all of the calls (make several
calls to each distant modem).

Then take a look at the connect speeds which you got on those
calls.  Throw out any odd connects that were either high or low
compared to everything else.

If you get 16K connects, your phone is just about a bare minimum
spec'd line.  With any half decent line you will get better than
21K connects most of the time.  If you can get 24-26K connects
on a regular basis, your line is about average.  If you get
28.8K connections you should whisper your thanks for a very fine
line... and if you often get better than that just be very glad
for your good fortune!

Of course all of that is relative...  If you live across the
street from the telco office and can only get 26K connects then
you have a really bad example of a telephone line.  And if you
are, like I am, on a cable carrier and 4 miles from the remote
unit, then 26K means your line is as good as can be expected!


Floyd L. Davidson          Salcha, Alaska

From: (Floyd Davidson)
Subject: Re: Advice on dealing with Ameritech
Date: Sat, 06 Sep 1997 05:30:13 GMT

Hunter Chisholm  <> wrote:
>We live in rural Illinois and has always had hum on our telco line. All
>utils are underground. Other houses in the area, and the same street do
>not have hum like we do. I've called Ameritech several times to report
>it, especially since it sometimes goes completely dead (especially when
>it rains a lot). Everytime they come out the "fix" (change pairs) the
>line but the hum never goes away. They say it is within spec, I say it
>is annoying, distracting, and we should have as good of phone service as
>the folks next door. Upon pressing the issue with the line man he says
>it is a problem. He has no answer when asked about a solution.
>I've tested my side the network interface box with a high quality butt
>set and there is still hum so I know it is not within the house wiring.
>Any comments on how to get quieter lines would be appreciated.

Your description certainly indicates a problem.  Technically the
requirement is that your line have a signal to noise ratio of at
least 24 dB.  The actual noise level might vary a little, as
will the signal level, depending on how much cable is involved.
Generally though, a 24 dB SNR means that you can barely hear
noise on a standard telephone set if you have good ears and know
what to listen for.  If you are being annoyed by a hum it isn't
likely to actually be within specs.

However, note that if you have a 25 dB SNR, which clearly is
good enough to meet tariffed specifications, you will have a
great deal of difficulty getting good connections using a v.34
modem!  You may never get better 16k or 21k on a good day.

Hence you do want to find a way to get the problem fixed.
Generally with little things like a bit of annoying noise, where
the phone does still work for voice calls even if they aren't
high quality, it can be frustrating to work with the telco.  The
method which works best is to regularly, and politely, file
trouble reports.  Keep track of who you talk to.  Keep a log
which lists the time and the name or initials of the person you
spoke with and a very brief one line description of what was
discussed, specifically including any expected action and any
specific numbers, names, times, etc.  A contact name, telephone
number, and time is good.  Ask them specifically for their
trouble ticket number.  And be aware that they are keeping an
identical log too, so being very obvious that you are logging
the information will not seem at all odd.  And don't be upset
if several people seem to be tossing it back and forth and
each one needs a totally new description.

Do not pick on the poor clerk that answers the trouble report
line!  That person is not a technical expert, and if they were
they wouldn't be wasting time answering unknown telephone calls.
When you do get a technician or a management representative on
the horn it is OK to be a little cranky and a little pushy, but
don't overdo it.  Direct questions like "What are the specifications
and how does my line measure up?", or "When can you actually fix
this thing?" are OK.

If they close the ticket, wait a day or two and open another
one.  It won't take long before someone decides that giving you
better service is cheaper than responding once a week to your
trouble calls.  The effort and expense they will put into fixing
the problem goes up dramatically when they begin to recognize
your name or voice...


     Floyd L. Davidson                           Salcha, Alaska

From: Floyd Davidson <>
Subject: Re: Loop Loss vs Frequency?
Organization: University of Alaska Institute of Marine Science
Date: Wed, 31 Jul 1991 12:24:12 GMT

In article <>
(Nick Sayer) writes:

> There are, as has been mentioned already, certain standards a POTS
> line must meet. Among them are:

>    low-end cut off
>    high-end cut off
>    frequency equalization
>    signal-to-noise

> If I am not mistaken (and I probably am), a standard POTS line must be
> like +/- 3 db continuously from 250 Hz to 2500 Hz.  If they're
> notching at 2 kHz, they're violating the tariff.

Maybe somebody who works around a line switcher can give some better
figures.  I searched around a bit, and coudn't find a definitive spec.
However, in the "Notes on the Network - 1980" book (and I could NOT
find it in the '90 version) it says something to the effect that an
end to end dial up circuit should have (this is NOT *must* have) no
more than 14 dB rolloff at 404 or 2804 Hz compared to 1004 Hz, and the
distribution would be 3 dB per each subscriber loop and 8 dB for CO,
mux, trunks etc.

That is not +/- but rolloff as they are not even considering the idea
that it might be less loss than at 1004 Hz.

> Paying for line conditioning doesn't get you much beyond a guarantee
> that they're meeting the spec they're supposed to be meeting in the
> first place.

It *might* guarantee you will get a circuit that is as good as
"normal", which is no where near as bad as it could be and still meet
specs.  I'm not positive about dial up lines, but I believe that the
only spec that actually changes is instead of 26 dB s/n you get at
least 28 dB s/n.  The things that might actually be causing problems,
like impulse noise, are exactly the same.  It is mostly a case of they
probably will make an effort to find the best facilities (like trying
different cable pairs etc.) they can.

> Your repairman should have a box that will measure each spec, and you
> should have a copy of the tariff. You should be able to watch as he
> plugs the various devices into the protector and reports the results.
> If any one of them doesn't meet tariff, you will get satisfaction or
> you can call the PUC and have the BOC roasted over an open fire.

Geeze, *I* don't even have a tariff (not one for my tele line anyway,
and it would take me a while to locate one on the trunks I test).

It must vary from one place to another, and may be different for local
lines, but in my particular case you can ask for and get a signed copy
of the data on a leased line that I do the testing on.  You can watch
if you keep quiet too!  And you will pay the initial installation fee
every time I do it too, so don't get wound up and ask to get it
checked every month.  (For C2 and/or C5 data conditioning on leased
lines it might be more than a couple hundred bucks.)

Floyd L. Davidson   | Alascom, Inc. pays me, |UA Fairbanks Institute of Marine| but not for opinions.  |Science suffers me as a guest.

From: (Floyd Davidson)
Newsgroups: alt.dcom.telecom
Subject: Re: POTS Frequency range
Date: 8 Jul 1997 17:59:01 GMT

In article <>,
Gary Gorton  <> wrote:
>300Hz to 3400Hz
>David Haid wrote:
>> Does someone out there know the Freq range of POTS in hz.???

The guaranteed minimum is 300-2800Hz.  The absolute maximum that
can happen is about 80-3750Hz.  Probably the average is about the
300-3400Hz range.


Floyd L. Davidson          Salcha, Alaska

From: (Floyd Davidson)
Subject: Re: Testing 2 wire lines in field
Date: Sun, 14 Dec 1997 05:11:48 GMT

<> wrote:
>John >Dearing wrote:
>snip .. snip
>> A really good piece of "test equipment" for finding telephone network
>> impairments is an old 2400 bps modem. No MNP, no error correction, no
>> nuthin'. JUst 2400 bps. Use the modem to make a connection to another
>> modem. Then sit and watch the screen. If you start to see repeating series
>> of "garbage" characters, you probably have a problem in the network. This
>> can be caused by network timing errors known as "clock slips". They appear
>> to the modem as large phase shifts of the signal. Since 2400 bps encoding
>> uses the phase of the signal to transfer the data, the modem try to decode
>> the shift and you keep getting these seemingly random characters. Usually
>> it's a series of {{{{{'s or sometimes the square root symbol. It all
>> depends.
>> I can't tell you the number of times that my old 2400bps modem has proven
>> trouble even when the expensive test gear didn't show any impairments on
>> the line.
>snip .. snip
>I concur.  I maintain sites with high speed radar data connections, but
>the slow 2400 terminal connection will show system errors before the
>other data connections.  It is very handy for spotting telco problems.
>I wish MCI had the expertise to use this method :(

But all it shows is _one_ basic error, and virtually nothing
else.  A 2400 bps modem is probably the best tool around for
detecting controled clock slips, but it won't give even a hint
of any other specific problem.  Ten years ago, or even five,
when digital switching and transmission facilities were new to
the many telco techs it was exceedingly common to see sync
problems, but today it is no more (or less) common than any
other type of problem.  And today there are very few techs that
have not dealt with it at least once.

A good v.34 modem, e.g. a USR Courier, can give a great deal of
info about SNR and bandwidth, but the values are all relative
and cannot be compared to values from any other modem or measuring
set because they lack accuracy.

There is just no substitute for a Transmission Impairment
Measuring Set (TIMS) which can produce reliable values for
amplitude distortion, SNR or noise with tone, idle channel
noise, impulse noise and phase hits.  Unfortunately for the
technically inclined customer there is almost no way to use
such test equipment without cooperation from the telco, and
they usually are not so inclined...


Floyd L. Davidson   <>   Salcha, Alaska

From: (Floyd Davidson)
Newsgroups: comp.dcom.modems
Subject: Re: Trouble Dealing with a Telco...any advice?
Date: 19 Aug 1999 06:20:59 GMT

Brian <> wrote:
>John Navas wrote ...
>> The 26.4 Kbps speed suggests that the customer is served by a
>> non-integrated DLC (aka SLC or Pair Gain).  If so, there is no chance
>> for higher speed unless the telco can be persuaded and is able to
>> switch the service to pure copper.
>Guess I'm the case that says the opposite.  There is indeed a chance,
>if things are done correctly.   I am on a known and admitted to Pair Gain
>line.  Here is one of my 3COM diagnostic reports:

You are the proof that he _doesn't_ have any chance at all.

There are two ways to engineer Subscriber Line Carrier (DLC,
SLC, pair gain, whatever one wants to call it), and the above
indicated "non-integrated" variation will result in v.34
connects at 24k to 28K, but usually right at 26.4K.  The
integrated variation results in the exact same results as using
a "pure copper" loop, which if the loop is short enough will
result in 40-50K v.90 connects.

But 26.4K connections are an almost sure indication that the
line is indeed on a non-integrated device, and pulling the telco
manager's teeth with gas pliers and no pain killer would be
easier than having it re-engineered and installed as an
integrated unit.


Floyd L. Davidson                
Ukpeagvik (Barrow, Alaska)

From: (Floyd Davidson)
Subject: Re: Lost modem speed when Pac Bell split my line
Date: 11 Dec 1999 12:44:02 GMT

tom  <> wrote:
> (Floyd Davidson) wrote:
>> I _guarantee_ you that a line which gets 26.4K v.34
>> connections is going to meet telco specs by a rather wide
>> margin!
>I had been told by Bell Atlantic Tech support that FCC defined
>minimum parameters made 26.4K the minimum acceptable baud rate
>on analog.  Does the FCC has a different minimum standard?

Actually there is no FCC standard for local telephone lines;
however, FCC standards are the same as others (ITU, IEEE, etc.),
and 14.4kbps is the right bit rate, not 26.4K.  I have a couple
of lines here that get 21.6K, and I can assure you they meet all
specifications.  (AT&T pays me to make sure of it.)  They are
both Foreign Exchange lines, one of them on compressed digital
facilities and one on old analog facilities (both are satellite
circuits and so are the facilities).

>My experience with SLiC is unique.  The noise was always one
>way - going out. Demonstarted the fault to a BA Tech using a
>2400 baud modem.  Suddenly the tech went silent, then agreed to
>execute some changes.  The line was perfect with no noise until
>I used AT commands to force a constant 2400 baud tone on one
>line.  It appeared on the other line with random, short noise
>spikes quite audible but not there until the constant tone was
>put on the line.  They replaced the fiber optic terminators in
>the CO and SLiC.  Suddenly neighbor's fax problems stopped

Heh heh, you bit em where it counts.  I can't be sure from the
description, but it certainly sounds like they had a typical
clock sync problem.  If the clocks do not maintain sync, the
received data on a T1 line will either be too fast or too slow,
if it comes in to fast a buffer overflow will occur and the
solution is dump an entire frame of data and set the clock back.
If the receive is too slow the buffer will underflow, so the
clock is skipped ahead and an entire frame buffer full of data
is used twice.  Each frame is 1/8000th of a second, which is the
sampling rate.  So each circuit takes a phase hit of that
proportion.  Your ears can't hear phase hits, so it won't bother
a voice conversation at all.  It isn't a big enough hit to
bother 1200 baud modems.  v.32 and v.34 modems have error
correction.  But a 2400 bps modem with no error correction will
garble every time. Here is a sample (a real sample too, saved
almost a decade ago just to demonstrate this),

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

Note the repeated combinations.  There are lots of '{{', but
there are also '{r', 'w3', 'xD' and others.  Noise does not
normally repeat in patterns, but clock slips do.  That makes it
easy to spot.

I can't tell exactly what you are saying you did.  It sounds
like you connected to a distant modem, and could hear hits in it
coming back?  What was probably happening is the hits were
really in the direction from you towards the distant end, and
the distant end is echoing the garble back to you.

There are some other possibilities given your description
though.  The above describes "controlled clock slips", but that
requires a buffer large enough to hold two whole frames.
Sometimes they don't use that big a buffer, and they can't drop
or repeat an entire frame.  So it just drops the connection and
resyncs from a cold start!  That puts some kind of a *major*
signal dropout on the line, loud enough to be heard and most
likely will knock the modem connection down every time.

Anyway, you stumbled onto the absolutely only way you can spot a
controlled clock slip.  Good show!

>  What is the difference between an non-integrated SLC and other SLCs?

The remote end of a SLC has a line card, just like one that
would normally be at the switching office.  Your cable pair
connects directly to it, and it has a codec that converts analog
to digital for the transmit (your voice) and digital to analog
for the receive.  A very nice cheap little device.  The side
towards the switch is Pulse Code Modulation (PCM), which is a
digital data stream with 255 discrete level values that are
encoded.  Those are sent to the switch via T1 carrier of some
kind.  It might be on fiber, it might be on copper... makes
no difference.  What counts is what they do with it at the

If they convert it back to analog (by passing it through another
codec) it can be hooked to a line card on the switch, which will
think it has an analog telephone connected to it.  That line
card of course also has a codec in it, and converts it back to
digital for switching and transmission to wherever you might
connect.  That is called a "non-integrated SLC", or maybe a
"universal" SLC.  It has the advantage that if the switch
already has many line modules, it is cheaper than any other way
to add an SLC.  It also looks and acts like what old time
engineers and techs are used to, so they tend to choose it over
an integrated system just because of ignorance.

The "right" way to do it in an Internet aware world, is to
connect the T1 carrier to a digital line module instead of an
analog line module.  That means it is never converted back to
analog, but goes straight into the switch the same way that
trunks from other switches do (which makes it different than
what they are used to).  One way they have to purchase a
"channel bank" to convert to analog and they must have or
purchase a line module.  The integrated way requires only a
digital interface (it might even be called a trunk interface,
which confuses folks too) be installed.

The difference to you is that extra conversion to analog and
then again back to digital.  Each set of those adds noise to the
signal.  That is because there are only 255 possible discrete
values, and in real life the actual values can be in between
those 255 that are allowed.  The difference between what it
really was, and the approximation that comes out the distant
end, is noise.  It is called quantizing noise.  It limits the
signal to noise ratio to 37 dB at best.

But a v.90 connection only has 1 codec in the circuit!  That
one on your line is there.  But at the ISP there is no codec.
There are no modem "tones" that get modulated.  Instead raw
data bytes are plugged into the PCM channel directly.  A byte
value of 100 is a byte value of exactly 100, and what comes
out the other end is not an approximation!  Hence no quantizing
noise.  Except if you are on a non-integrated SLC with those two
extra codecs.  They put way too much noise into the line to
allow v.90 to work.

Hence you end up using v.34, for which there is an original
modem tone that is modulated and then encoded.  Two PCM channels
back to back will both have the same quantizing noise, so the
SNR is reduced by 3 dB (two equal signals added together is a 3
dB change), the max you can get is 34 dB.  Add the random phase
noise from those two codecs, line noise etc., the additional
frequency rolloff from going through two extra line card
filters, and bingo you can only get 26.4K with a v.34 modem.

Since you were getting an odd 28.8K connection, it is also
possible that you are on an integrated unit, and are at the
limit of cable length that can function well at 28.8 on an older
modem, but will on a newer modem.  (I was on a line like that
for several years.  Now I have lines that are not on a SLC, but
they are too long for v.90, so I get 28.8K and 31K connects.)


Floyd L. Davidson                
Ukpeagvik (Barrow, Alaska)

From: Floyd Davidson <>
Newsgroups: comp.dcom.modems
Subject: Re: 3.4KB/sec is the maximum on my USR Sportster 33.6K Faxmodem 
Date: 04 Jun 2000 12:51:55 -0800

Daveman_750 <> wrote:
>If you consistently get V.34 connects of 28800 and higher and
>it stays there, V.90 is virtually guaranteed to work.

That is not true.  The two protocols have very distinctly
different requirements and getting good performance from a v.34
modem is NOT an indication that a v.90 will even be able to

I can provide two very typical examples of when that happens as
both my home and my work locations are just the right distance
from the telco that no modem will negociate a v.90 connection,
but any half decent v.34 modem will get a minimum of 28.8Kbps.
The problem is that I am right at 3 cable miles from the telco
in both cases.  Very annoying... :-)

These are not rare cases, but just a demonstration of how
different cable characteristics affect the two protocols

>  If you had a lot of analog noise, you couldn't get 28800 and
>more than 1 a/d conversion wound usually only allow 24000 to
>26400.  However, if you have bad digital impairments, you may
>not get good 56k.

Generally (and the number of codecs is an exception) there is no
difference between what you are referring to as "analog noise"
and "digital impairments".  In both cases it is characteristic
of the cable.

>  The performance difference is HUGE if you do more than just
>read e-mail and you can get a GOOD V.90 connection around 44K
>or higher.  I would borrow a friend's 56K modem and if you get
>44K or higher and the throughput is matching or near-matching,
>buy a 56K modem.

Despite the mistaken impressions above, the overall point is
*exactly* what is said here!  This is _good_ advice which should
be taken.  If a v.34 modem commonly gets a connection at
anything above 21.6kbps or so, the only way to know for sure if
a v.90 modem will get a better connection is to try it.
Moreover, try it with with more than one modem.  (For example, I
have a brand name PCMCIA modem that simply will not make *any*
connection over typical satellite circuits used to village here
in Alaska.  Hence I pack around an old faithful SupraFAXmodem to
use with my laptop when I travel.)

>  Here in central NJ 56K modems have come down in price.  I
>have seen them for 5 bucks on sale after rebate.  No kidding, 5
> Linepithema humile <> wrote:
>> Addendum: Also when I connect, I get from 26400 to 28800. I
>> used to get 26400s often, but I get 28800 often these days
>> and get 3.4KB/sec during good hours instead of 2.8KB/sec. I
>> know initial connection speed numbers change often when
>> connected. :)
>> Linepithema humile <> wrote:
>> > Is 3.4KB/sec considered maxxing out my old modem? I used to
>> > get 2.8KB/sec but somehow my speed was improved the last
>> > few weeks [maybe GTE did something [grin]]. I wonder if I
>> > should upgrade to a 56K modem. I was downloading MP3s,
>> > compressed files, MPG videos, etc.
>> > What do you guys think? I look forward to receiving replies. :)

For compressed files, that is just fine for a v.34 connection.  As
advised above, see if you can borrow a friends v.90 modem or find one
cheap enough to buy at throw away prices, and give it a whirl.


Floyd L. Davidson                
Ukpeagvik (Barrow, Alaska)

Index Home About Blog