[Asterisk-Users] New $89 VOIP phone

Chris Shaw chriss at watertech.com
Wed Aug 18 13:44:44 MST 2004


If anyone wants to discuss my reasons please message me off list, this is
getting ridiculous... All I was trying to say is that if I was bidding a job
I wouldn't even think of suggesting splitting or 900' cable runs unless the
customer really knew what they were asking for... There are some switches
out there (mostly older ones grant you) believe it or not that stick quite
firmly to TIA and ANSI specs and don't like extra long cable runs or
excessive noise... It can be kind of annoying when you're going to all the
trouble of wiring the premises and splitting cable and what not to find that
the customer's switch doesn't play nice with your half-assed non-spec
cabling scheme... Now they not only have to pay you for your time, but they
have to buy new equipment...

What's worse is you probably won't find this out right away, but later on
when the customer complains that they have performance issues, etc... Not to
mention the fact that if the customer EVER wants to use Gig-Ethernet, forget
it...

As far as data corruption goes, YES on the wire, data can become
corrupted... Fortunately for us, there's TCP and to a lesser extent UDP
which inspects the packet and decides through check-sum that it's not the
right size... Guess what happens then? The packet has to be resent adding
latency to the network... If this keeps happening, TCP/IP will begin to back
off, lowering throughput... This is BAD for VoIP...

This is all I will say on the matter... This is getting WAY OT... I
shouldn't have said anything in the first place, I apologize...

    -Chris

----- Original Message -----
From: <asterisk at bsius.com>
To: <asterisk-users at lists.digium.com>
Sent: Wednesday, August 18, 2004 12:04 PM
Subject: RE: [Asterisk-Users] New $89 VOIP phone


> Chris Shaw [chriss at watertech.com] said:
>
> > You're kidding right?
> >
> > There's a reason why category 5 cable is twisted the way it is... to
> eliminate
> > or greatly reduce RF crosstalk on the wires... Now what would happen if
> you
> > split those wires between 2 different signals and kept them tightly
packed
>
> > together?
> >
> > Can we say random data corruption, mysterious errors and terrible
> performance?
> > Bingo!
> >
> > Again, this kind of thing might be ok for a small home network, but you
> can't
> > seriously suggest it for a business...
>
> Usually generic cabling standards are used to specify twisted pair cabling
> for a commercial building. These generic cabling standards are TIA/EIA
568-A
> and 568-B (ATT 258A) in the US and ISO/IEC 11801 internationally. These
> standards generally specify 4-wire pairs per connection.
>
> Those standards do not attempt to address what kind of signals will be run
> on the different categories of cable (cat 3-7 for TIA/EIA and Class A-F).
It
> is the responsibility of the applications standards group IEEE 802.3, ATM,
> etc... These groups tend to use existing generic cabling standards so as
not
> to create their own and have to promote these new standards.
>
> With that being said, as 100BASE-TX uses only 2 out of the 4 available
pair
> it is entirely feasible to utilize the remaining 2 pair for traditional
> analog teleco, or another 100BASE-TX signal. This could limit your ability
> to expand to newer technology that may require all 4 pairs.
>
> Unused wire pairs can affect the crosstalk from the 1,2 to the 3,6 wire
> pairs (orange/green in 568B) that are used in 2-wire pair Ethernet
slightly.
> They can have a larger influence on external noise susceptibility and
> emission of LAN signals on the cable into the air. The unused wire pairs
do
> function as a poor shield, but not enough to cause major heartache. And
> according to ANSI/TIA/EIA 568-B, when terminating jacks all pairs must be
> twisted to within 13mm to the point of termination. I've seen very few
> installations where this is followed, most ever time these test out fine
for
> NEXT, though there is no reason not to wire things according to the
> standard.
>
> I doubt though, that you'll find this enough to cause any major issues
with
> the Ethernet signals, especially if you're under 100', crosstalk becomes
> more of an issue the father the signal has to travel.
>
> In any case, the quality of the transmission is as much, if not more,
> dependant on the Ethernet transceiver as it is the signal line. Poor
quality
> Ethernet transceivers will cause as many, if not more, problems as poor
> quality line. Take, for instance, Cirrus Logic's CS8952T which is capable
of
> transmitting well over 160M on CAT-5 in a "noisy" EMI/RF industrial
> environment.
>
> As far as your statement about random data corruption, mysterious errors
and
> terrible performance. Data corruption should never be an issue when you're
> using protocols that provide error correction (IP). If using UDP, most all
> UDP applications provide some sort of error correction or discarding of
> invalid packets, since UDP should always be used in instances where data
> integrity is not important this should be of little concern.
>
> FWIW this type of situation can be found at a lot of universities (two
> 100BASE-TX signals on a single CAT-5 cable. Though most, if not all, have
> stopped this practice.
>
> Personally, I currently have several installations where I am running two
> Ethernet signals over a single CAT-5 cable (all 4 pair), through testing
has
> uncovered no ill effects of this installation. We have also, in the past,
> been forced to utilize existing CAT-3 cabling where it was "impossible" to
> run certified cable. The existing cable tested out properly and has been
> running trouble free ever since.
>
> As far as suggesting it for a "business" environment, probably not. When
> possible it's always best to run the best (or better) cable for the
> application. However, as we all know, budget or physical limitations may
get
> in the way. If this is the case, just make it clear up front of the
possible
> issues that could be encountered and establish the procedures and costs
> associated with remedying the issues if they do arise.
>
>
> With all of that out of the way, if you are putting two 100BASE-TX signals
> over a single CAT-5 cable, it should be obvious that you will not be able
to
> use PoE from the head end. I've never tried it, but I would imagine that
> your equipment would not care too much for the 48VDC on it's signal lines
> and it more than likely would not work with the 802.3af standard anyway.
> But, if you can afford the PoE switches/injectors in the first place,
> running another CAT-5 drop should really be the least of your worries. You
> can still use a PoE injector just before the device itself, or use a wall
> wart transformer.
>
> As far as the suggestion of using Ethernet cable splitter. This is
possible,
> however I've seen many products that claimed to be splitters but were
> actually hubs or something to that effect. IMHO they are all a waste of
> money, because an additional keystone jack is under a dollar and would
make
> a tidier installation in the long run anyway.
>
> If it were me wiring this, I would break out pins 1-2 and 3-6
(orange/white
> - orange and green/white - green) to the same corresponding pins on jack
1,
> standard 568B.
>
> Then pins 4-5 and 7-8 (blue - blue/white and brown/white - brown) to pins
> 2-1 and 3-6 (orange - orange/white and  green/white - green) on jack 2.
>
> Crude ASCII ART follows (make sure you're using a monospaced font to view
> this):
>
>   JACK 1   1  2  3  4  5  6  7  8
>   PC #1    |  |  |        |
>            |  |  |        |
>            |  |  |        |
>            |  |  |        |
>            |  |  |        |
>            |  |  |        |
> 568B Port  1  2  3  4  5  6  7  8
>   Wall              |  |     |  |
>                     |  |     |  |
>               +-----+  |     |  |
>               |        |     |  |
>            +--|--------+     |  |
>            |  |  +-----------+  |
>            |  |  |        +-----+
>            |  |  |        |
>            |  |  |        |
>            |  |  |        |
>   JACK 2   1  2  3  4  5  6  7  8
>   PC #2
>
> -Bill
>
> _______________________________________________
> Asterisk-Users mailing list
> Asterisk-Users at lists.digium.com
> http://lists.digium.com/mailman/listinfo/asterisk-users
> To UNSUBSCRIBE or update options visit:
>    http://lists.digium.com/mailman/listinfo/asterisk-users




More information about the asterisk-users mailing list