[CLUE-Tech] CLUE: DSL - PCISYS

[Timothy C. Klein]

* Randy Arabie (rrarabie at arabie.org) wrote:
> On Wed, 3 Jul 2002, Timothy C. Klein wrote:
> 
> > Those would just amount to really short bridge-taps.  DSL is spec-ed to
> > allow something like 1500 or 2000 feet of bridge tap and be OK.  Doesn't
> > seem like there would be enough length for the signal to reflect, get
> > out of phase, and cancel signal when it comes back.  Is you your house
> > *really* big? :-)  Or am I missing something.  Just curious.
>  
> Sounds like you know a lot more about it than I do...I don't have any idea 
> what a bridge tap is.

Bridge tap is when your telephone line has a long length that does not
go to your house.  It is hard to describe with out a picture.

CO=|----------------------------------|=House
         \
          \
           ------------------|.Bridge Tap

The bridge tap does not actually connect to a phone (usually), but it is
terminated somewhere.  The DSL signal, when it reaches the split, takes
both paths.  When it his the end of the bridge tap, it bounces back.
When it is coming back wards, it gets out of phase with the rest of the
signal.  If the length is right, the out-of-phase will be enough to
start canceling parts of the signal that were coming from the CO --
thus you run into trouble.  The length that becomes a problem depends on
the frequency of the signal.  For POTS, that signal frequency is low,
and the length of wire needed ends up being very large.  That is why the
phone company lays wire this way -- bridge-tap is not a problem problem
for POTS, and it gets them a little bit of flexibility in their cable
route.  DSL, though, is much higher frequency, and thus is subject to
harmful interference at much shorter lengths of bridge-tap.  I haven't
done this stuff for a while now, so I hope I am remembering it correctly.
I am working on a Physics degree, so technically this stuff is all
within the realm of my abilities, but tracking down the numbers would be
a big pain :-)

> No, my house isn't very large.  The run from the 'demarc box' where my
> phone line comes inside to where I have my CISCO 678 is only 20'.
> However, there is some 'old' phone line for a little ways, then a
> splice with some new line, then another splice which terminates where
> I have my CISCO 678.
> 
> I've read, and been told, that each of those splices degrades the
> quality of my signal and *could* be the cause of my problems.  I read
> that on some website a few months back....can't recall the URL just
> now.  And, I was talking with a guy (who is NOT a DSL authority) that
> had a 'professional' installation. He said that was the recommended
> method (i.e. a splitter at the 'demarc box' and a dedicated line to
> the DSL 'modem').

I guess it is possible what you are describing could cause problems,
especially if the line was marginal to begin with.  It would not,
however, be caused by signal reflection.  The length is just too short.
I guess it would have to be caused by poor splices.  I am skeptical that
it would make a huge difference, but it might, who knows.  I know that
my 100 year old house has some very 'interesting' phone wiring, but my
connection speeds are just fine.  

> I can't vouch for the accuracy of that info, but it sounded plausible to me.

It would be interesting to see if you get a speed jump by changing it.
Qwest uses splitter-less installations, and instead goes for those silly
little filter-things.  I thought the splitter was really just to reduce
noise on your *POTS* line, so it didn't really matter.  I could be
wrong.

Tim
--
==============================================
== Timothy Klein || teece at silverklein.net   ==
== ---------------------------------------- ==
== "Hello, World" 17 Errors, 31 Warnings... ==
==============================================