Index Home About Blog
Newsgroups: comp.dcom.modems
From: (Floyd Davidson)
Subject: Re: CAT 3 cabling
Date: Sun, 23 Nov 1997 10:26:19 GMT

I've done some reformating to make reading easier, and hopefully
I haven't cut too much of the context out...

Art Jackson  <> wrote:
>Kent Salzman wrote:
>> Jack McClain wrote:
>>>IMHO the use of CAT5 or "twisted pairs" is unnecessary and
>>>improves nothing over standard "inhouse" telephone wiring
>>>UNLESS, more than one computer is being used in the same
>>>cable. The advantage of cable with pairs that are

In that case, yes it is true, UNLESS there are other sources of
"noise", which there may or may not be.  One that is rarely
realized to exist would be odd harmonics of the power line 60Hz
frequency. Other sources are computer network cables, RS-232
cables, computer equipment, arcing of switch contacts of power
equipment, and many others.  So it might not be a case of
running wire for what exist today, but more on the line of
wiring for a future use to avoid unknown future noise sources.

That said, CAT5 has absolutely no technical advantage over CAT3
at voice frequencies.

>>>individually twisted is in an environment where there are
>>>more than one computer using separate pairs in the same
>>>cable, that are in use at the same time. Then individually
>>>twisted pairs helps eliminate cross-talk, or bleed-over,
>>>from one pair to the other.

CAT3 is also twisted pair.  And at voice frequencies will eliminate
cross-talk and other noise induction.

>>      That is in fact the major reason for using multiple-pair twisted
>> cable for computer networking.

The significant differences between one or another type of
twisted pair have to do with the twist.  A shorter twist works
at higher frequencies, but it also makes the cable physically
larger and requires more copper per foot of cable too.

Generally speaking longer twist cable is useful for large multipair
cables (typical telephone 600 pair cable, for example), and shorter
twist cable is _required_ for higher frequencies (10bastT ethernet,
for example).

Hence, if you are positive that the wire you run will never be
used for anything other voice frequency (0-4Khz) telephone use,
use CAT3 if it is significantly cheaper.  However, balance that
against the chance of ISDN, xDSL, or 10BaseT use at some point
in the future, and it might well be much better, and cheaper, to
go with CAT5 the first time.  That is particularly true if you
pay wages to have the work done.

>>     However, the stuff does also have, as it needs to have, low
>> electrical impedance.  Impedance in a circuit includes the product of
>> the unit impedance of the conductors and their length.  If there is too
>> much total impedance, high-frequency signals are attenuated which, in
>> the case of modem wiring, can result in lower-speed connections.

That is nice sounding gobble-D-gook, but it isn't true.  The
characteristic impedance of twisted pair cable depends on a
number of factors.  Typical low twist 26 guage plastic insulated
telephone cable runs up around 1200 ohms.  Typical high twist
ABAM used for DS-1 cabling is close to 100 ohms.  Cable with
loading coils will also have a lower impedance, but obviously
will not have better high frequency characteristics.  In any
case, the impedance has little to do with frequency response at
voice frequencies used by modems.  High frequency rolloff is
caused by parallel capacitive reactance more than probably any
other characteristic.  Small wire and long runs...  mean high
frequency rolloff.

>>     How long is "long"?  That depends on the wire that's in place, so I
>> can't tell you.  However, I will be replacing a temporary fifty-foot
>> flat cord with twisted permanent wiring shortly and will be interested

That is correct to a point.  How long is long does depend on the
wire.  For flat cord (which is NOT twisted), anything over 6
feet is long, and sometimes even 6 feet is too long!  For
twisted pair designed for telephone voice frequency applications
(e.g., CAT3) the effects of distance do not become apparent for
2-3 miles.  At 5 miles significant effects can be measured, and
may be a problem (v.32 is fine, but v.34 will likely not get
28.8Kbps connections).  At 10 miles significant difficulties of
several kinds are guaranteed.

>>     I wouldn't put off new shoes to install twisted-pair.  But anyone
>> who thinks it might help his particular situation and doesn't find the
>> idea impractical might just as well go ahead and do it.  He'll feel
>> better--and it can't hurt.

Very good advice.

>FWIW, I'll add my recent experience with wiring upgrade here...
>...The outside drop wire was a single pair non twisted wire...

Which can be identified as "horrible stuff"!  Replacing it is almost
guaranteed to be helpful in all but the most basic of configurations
(with only one line, it may not ever be noticed...)

>rewired the whole house using CAT 3 (2 pair, 24ga.)...
>...When the Telco installer came to put in
>the second line, she replaced my single pair non twisted drop with a two
>pair individually twisted pair drop...

Did you give her a tip?  :-)

>I would have to say that the biggest improvement in my case was the
>outside drop wire replacement that bumped my connect speeds up one

Good observation.  It probably did much more than you even
realize too!  What you have seen were merely the initial connect
speeds, and what that change has done might only have a slight
effect on the initial speed, while it may have eliminated many
many instances of intermitant random noise hits that later drop
your connect speed down to a rate that can be maintained by the
modems with less error correction required.

>Now, the total wiring from the Telco switch line terminals, all
>thru the cable pair, drop wire, and inside wire to my jack is CAT 3,
>individually twisted pair wiring and provides a fairly uniform impedance
>from end to end (including loading coils)...I am located about 3 1/2

Not likely to be including loading coils.  At 3 1/2 miles they would
be of marginal use as originally intended (though if most customers on
that cable are even farther down the line then loaded cable would
very likely be used at that distance).  But at 3 1/2 miles a normal
26H88 cable would have 3 sets of 88mH loading coils (every 6000 feet, that
is what the "H" in 26H88 means), and you would be hard pressed to get
a PCM modem to work on that cable, much less at anything like the
speeds you are indicating.

>miles from my Telco CO and have copper pairs all the way...Another case

Are you positive it is copper all the way for 3 1/2 miles????  I doubt
that, loading coils or no loading coils, you will get 50Kbps+ connects
over that much cable!  If it is happening, you are lucky!!!

>of wiring problems I recently experienced was at my Son's house in AL.
>I took a USR x2 external down to him and we installed it on his new P233
>computer...First call connected at 45.3K and several after that did too.
>Where his computer was set up, he had run a 25 ft. line cord around the
>end of the room to the phone jack on the opposite side of the

That much flat non-twisted pair line cord is enough to cause
audible noise problems on voice calls depending on the
environment!  I bet the actual data rate after a bit of time was
even worse than that!

>room...Since it was an external modem, I sat the modem out in the middle
>of the room, then connected it with the 6 ft. line cord that came with
>it...Connected at 50.6K every time...Wound up rearranging furniture to
>move the computer close to the phone jack but he is a happy camper with
>50.6K connects and occasional 52K...In his case, the 25 ft. line cord
>was eating up his connections...

That fits the pattern perfectly!

>| Art Jackson       W4TOY |   When all else       |       |
Hmmm...  want to understand twisted pair cable?  Read the Radio
Amateur's Handbook section on open wire transmission line.  Same
thing, except at lower frequencies!  The twist is because it can't
be kept physically away from everything far enough, so by twisting
it the influence on the two wires will be essentially the same for
anything that causes induction.


Floyd L. Davidson   <>   Salcha, Alaska

Index Home About Blog