how-to block ads
|
Bicephale
join:2005-09-24
 ·TekSavvy Solutions..
| The customer's own wiring
Filtering: Distributed vs Central |
In ADSL technology the customer is responsible for
his phone wiring until he reaches the "demarcation
point" set by the telephone company and that's one
aspect which is often neglected even if it's under
his very own control. Lets look closely at what's
involved here, precisely! Well, for starters, any
extra length of unsuitable wiring must be avoided:
AM radio and DSL signals overlap in the 535 KHz to
1104 KHz frequency band and it means interference;
ADSL2+ is exposed to more nuisance in the 1104 KHz
to 1605 KHz band... "Balanced" design combined to
twisted pair cabling can reduce the susceptibility
of DSL signal to external sources of noise while a
straight non-filtered cable looks like an antenna,
any eventual taps along its path complicating this
problem by adding impedance mismatches which would
cause more noise due to parasitic reflections. An
other significant problem to address is related to
possible interactions between the customer's phone
appliances and his DSL signal: some hissing sound
in the earpiece, disrupted browsing while dialing,
etc... As a result of all those issues "low-pass"
filtering is mandatory and central filtering would
be ideal: it reduces the parasitics to a minimum.
The 1st schematic helps to illustrate that even if
it's hidden behind walls the wiring contributes to
noise as long sections of it are allowed to behave
like antennas compared to the 2nd drawing where it
is clear that the "hot" (red) path is minimized...
Both configurations are built around the very same
components but only the last one reduces nuisances
optimally as it keeps the parasitics away starting
at the root; in one case, it seems a customer may
easily accumulate up to a hundred feet of poor DSL
wiring while in the other this can be brought down
to a few inches - no fancy twisted cable required.
Finally, let me show what i consider nearly ideal:
%20.PNG)
This setup accounts for the fact that some filters
aren't reversible: plugging a pair of these as my
drawing suggests will keep the shunting capacitors
away, reducing the risk of an unwanted interaction
in presence of an additional filtering device, for
example. The ground path which most cheap filters
happen to interrupt can be re-established when the
customer feels he has a good reason to require it.
Beyond that, i guess that an NID filter is better:
Wiring from demarc., BellVictim, 2007-Mar-17 | |
|  |
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| Centralized Filtering
An important parameter in case of high attenuation
is wiring, persons living in appartments are bound
to use "distributed" filtering while those who are
in a private home can have "centralized" filtering
if they choose to. This section refers to the 2nd
option, it hardly applies to appartment situations
unless the landlord happens to be cooperative. My
text is intended at those who can call Bell to get
an "NID" installed outside of the building and i'm
suggesting they do if it's not there already since
the best-case scenario one can dream of is when he
has the opportunity to add a filter/splitter right
at the heart of his NID box - then a CAT-3 twisted
pair cable (or better) to create a clean DSL path:
NID Filter/Splitter (The Yellow Module on Top)
Ideally, the internal circuit will look like this:
Alarm-compatible "POTS Splitter"
Shielding
Sometimes centralized filtering doesn't suffice in
presence of nuisances though, that's why i'm about
to describe how i ended up trying to control noise
caused by kitchen activity! I had been collecting
Error Rate (aka "noise") curves for a while when i
thought of superposing some daily records and then
i finally noticed there were quiet vs busy periods
emerging: noise was linked to human activities so
i got the idea to check if it didn't come from the
kitchen by turning off the wireless phone but that
failed; it felt clueless so i turned on the other
appliances by pure curiosity and bingo! At 1st, i
couldn't tell which peaks i caused so i decided to
do it repetitively, at the begining of every hour;
starting from there, most of the noise which i was
creating became easily identifyable as it detached
in terms of time/level from almost any background:
%20.PNG)
Man-made Noise (Graphic Generated using GNet BB0060B MoDem/Router & 'DMT v2.x' 3rd-Party Utility)
I already had implemented centralized filtering at
the demarcation point so it wasn't clear what more
could be done and most specially beyond of it. It
seemed an inspection of Bell's wiring was in order
and i knew i couldn't count on their low wage sub-
contracting staff to do it extensively, somehow...
It didn't take long before i finally put my finger
on it: decades ago Bell had installed flat wiring
closely to the electrical power panel; many moons
later someone must have completed this messy setup
by packing all of those loose ugly wires together:
Improper Phone Wiring Installation
My 1st action was to separate Bell's wiring but it
wasn't enough so i started to experiment with ways
to cure the rest through shielding. It wasn't too
clear what was required in order to shunt magnetic
and electric fields in this situation so i started
my work with two sections of strong 1" square iron
tubing... This was quite some involving challenge
which required heavy handcrafting - not to mention
the fact that Bell's line had to pass thru it all:
 %20&%20Home%20Lines%20.JPG) %20&%20Home%20Lines%20.JPG) %20&%202nd%20Ground%20Bloc%20.JPG)
Final Centralized Filtering & Indoors Shielding
Anyway, my shields worked but it still didn't stop
all of the noise so i went outdoors and discovered
that Bell's flat cable also happened to follow our
PVC Power conduit and passed under Hydro's counter
as if it wasn't bad enough. It was impractical to
add more iron tubes so i prefered sleeves instead:
Ferro-magnetic Metal Sleeves
Here are the same shielding sleeves once in place:
Outdoors Shielding
Although highly innovative, i think any technician
who can excercise his judgment should be satisfied
with this shielding approach and most specifically
the part where sleeves are used as these happen to
be non-intrusive and go off easily after a test...
In any case, this graphic feels like thought food:
%20.GIF)
Comparative Man-made Noise Curves
The final conclusion here is avoid close proximity
to power conduits at all costs if that's possible;
there are no more excuses now for Bell's employees
to layout a dry-line, etc., in such a lousy way!!!
N.B.:
The last graphic has a red bar reaching beyond 200
errors per hour, i didn't cause it so i can't tell
where it came from but this strongly suggests that
other noise sources were still an important factor
which shielding didn't appear to cure efficiently.
A loose (high-impedance) shunt across the line can
produce such a disruption. Actually, it turns out
that i do expericence disconnections due to a loud
noise sometimes and it's momentarily attenuated by
"hammer pulse dialing" number 9 a few dozen times!
 | |
| | | | 
xgmranti
@magma.ca | I don't mean to spam meaninglessly but this should be stickied. That is one hell of an informative post. | |
| | |
Bicephale
join:2005-09-24 | Re: The customer's own wiring Thank you! I had a FAQ in mind when i started this but i changed my mind...
 | |
| | | | | | | | |
bellunder
@teksavvy.com
| Re: The customer's own wiring said by Bicephale  :And now, the final touch... this looks cool. Does it really reduce a significant amount of noise on the phone line? | |
| | | | | | 
Guspaz
Guspaz
Premium,MVM
join:2001-11-05
Montreal, QC
·Colbanet
4 edits | Re: The customer's own wiring Most of Bicephale's changes seem to be focused on eliminating the 60hz AC interference, although I could be mistaken.
Bicephale: It looks like that Bell cabling is both pairs in one cable. Would you get less interference if you replaced it (at least the inside wiring) with shielded twisted cat5e or cat6 (or to get insane, cat7 cabling, which shields both the cable AND the individual strands within it, so double-shielded) cabling? Or would that have less shielding? | |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| 
RFI Control using Toroïdal Chokes |
There was a time when i tried to evaluate toroïdal
chokes but the interest was quite marginal; there
were very few posts showing measured results (this
wasn't exactly a motivating episode, i shall say).
I guess some people no longer followed my progress
after a while or i'd read about RFI traveling over
60 Hz Power A.C. curents too... I got switched to
6016/512 Kbps Interleaved since then, it was 5056/
800 Kbps FastPath before but i don't VoIP/Game/P2P
so i find this most appropriate considering that i
no longer notice problems anymore when i browse or
DownLoad. My CRC Error rate curves have fallen to
such a flatness state i stopped collecting them...
To be exact, i can only feel the additional speed!
Consequently, i'm in no position to evaluate these
little devices but i can reply this: if i'm asked
to express my opinion as to wether toroïdal chokes
are worth a try or not, euh... my comment will be
«take a good look at the above picture, there's my
answer»! Feel free to do the same, i've got these
for free by recycling old Power Supplies so it was
not like i had something to loose! The phone line
comes right from the pole and the only wiring that
i could still make better is the short one between
the white outlet (connected to the demarcation box
by a couples inches of wire) and the MoDem... I'd
like to try a twisted pair as it should look nice,
eventually, but i'll confess i got no need for it.
| |
| | | | | |
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| Re: The customer's own wiring Hi Excaliber,
I have a question for you! It seems like a lot of
users believe that they can just "ByPass" whatever
wiring problem they have by connecting their MoDem
at the demarcation box while the filters are still
distributed through the house. Some of us fail to
realize that a filter plugged into the wall-outlet
only filters what's connected to it but not ahead;
in other terms, it won't occur to some people that
distributed (local) filters effectively leave most
of the house wiring parasitics in place even after
they worked hard to move the DSL device near their
demarcation box in the basement. I'm not implying
that this is your case, i should admit that my own
situation once was even worst, actually. I find a
few guys didn't know how to improve the quality of
the wiring as well... BryanViper, for example, he
needed close to four years to have his cleaned up!
Poor BryanViper, i just can't miss the opportunity
to show how long it could take to get these things
fixed when no suitable help is provided! Well, my
question follows, sorry for the long intro: would
it be satisfying enough if you just moved the unit
downstairs but without moving the filters with it?
In other words, would you get the same improvement
if you only tried to "ByPass" your house wiring by
setting the MoDem ahead, near the demarcation box?
| |
| | | | excaliber
join:2007-04-18
Laval, QC
1 edit | Re: The customer's own wiring I had first tried it without the filters on the splitter but my SNRM at startup was at 13. Once I added the filters on the other jack of the splitter and reset the modem it jumped to 18.5 (comparable to having the modem directly plugged into the demarc without my house wiring), so I didn't bother running/testing without the filters on to see the amount of errors. | |
| boast
join:2005-09-03
Miami, FL
2 edits | the choke thing is interesting.
Just ripped open an old PSU and tried it. My SNR went from like a 6.8 to a 7.3, and 9 to 10 in upload.
thanks. | |
| | | | | |
|
%20.GIF "94754 bytes")
). I removed the filters from the only phone that was plugged in and no static on the line and my house alarm hasn't given me an error.