Text 315, 146 rader
Skriven 2004-08-04 10:18:58 av Peter Knapper (3:772/1.0)
    Kommentar till text 311 av Michiel van der Vlist (2:280/5555)
Ärende: Fido on the move via cell phone.
========================================
Hi Michiel,

 MvdV> All true. can't argue with the theory. But how does 
 MvdV> it work out in practise. Thing is, I have found 
 MvdV> limits for ADSL ranging form 2 to 8 km, but I can't 
 MvdV> find range limits for ISDN everywhere. That suggests 
 MvdV> it is not a problem in practise.

As the frequency spectrum required rises, the type of cable also has a bearing
on the issue, thinner cable = greater loss for higher frequencies. Older cable
may actually be be better off because its often a thicker gauge than the newer
cable, but being older you may have loading coils to contend with. 

Voice uses up to 3400Hz, ISDN uses up to about 168Khz, G.Lite uses up to 1Mhz
(I think, I am not too sure) and full rate ADSL uses up to 1.3Mhz.

In addtion, for ADSL there are various types of encoding used (most still call
it "Modulation", but its really a CODEC function, or enCODing/DECoding) for
ADSL. Its actually split into 256 discrete "channels" or mini-frequency ranges
and the data stream is "encoded" into the channels. EG: G.DMT is full speed
8Mb, while G.Lite (also called splitterless) is around 2-3Mb. The main
difference is because the frequency spectrums used are reduced for G.Lite. 

EG: I can configure my ADSL Router to Sync up using either method, on G.Lite I
get around 1.6Mb, while on G.DMT I get around 2.8Mb (I am actually quite some
cable distance from the Exchange). When my ADSL was first installed using G.DMT
I got only 1.9Mb and it used to re-train 10 or more times per day, however I
was able to locate an expert tech who increase that to 2.8Mb by adjusting the
exchange end of the circuit by increasing their output signal slightly. They
don't like doing this too often because it tends to limit the number of ADSL
connections that can service via that cable bundle, however as they had a
policy to guarantee 2Mb minimum, they considered the tweak as "necessary". My
connection now stays up for weeks at a time before it retrains...


 MvdV> Maybe I did not search hard enough.

There is quite a good ADSL FAQ that goes into all this, the "DSL" technology
environment has litterally dozens of encoding schemes in use, most are for very
specific cabling needs. In some areas of this city, the Telco's use DSL type
delivery for native WAN Data Services, but these use a Synchronous (same speed
up & down), rather than Asynchronous form of DSL.


 > The biggest problem with signaling such as ADSL is at the
 > exchange where all the copper cables come close together, they
 > They have a really hard time stoping the signals from
 > interfereing with each other...

 MvdV> Indeed. But that is ADSL...

The same applies to Voice (POTS) and ISDN, but to a much lesser extent. Have
you ever heard a neighbours conversation on your phone line? Thats usually
because of a low resistance link (or even a dead short) between 2 adjacent
pairs caused by something like water in the cable, etc. Its much rarer these
days than it used to be, but it still can happen. Such "cross-talk" usually
kills ADSL stone dead!


 >  MvdV> Even if you are right, it does not apply if I
 >  MvdV> correctly understood Carol. She was referring to the

 >  MvdV> wiring in the apartment building wasn't she? Surely
 >  MvdV> that short stretch can't make the difference?

No, probably not in those terms, its more the cabling in the street that is
likely to be the issue. However if a "filter" is used, it really needs to be as
close as possible to the device it is filtering for, otherwise excess cable
lengths can come into play. In this case these are likely to be large numbers
so I doubt its the reason here...


 > > 2:  Cables can not carry more than a certain number of ISDN
 > > lines; don't ask me how many, I know that the limit exists.

 >  MvdV> Again: are you sure you are not confused with ADSL?

 > No, the signal loss and interference issues affects all
 > transmission types over copper lines.

 MvdV> Again: in theory. How is it in practise?
              
VERY very real. Near End Cross Talk (NEXT) is the single biggest issue to the
tranmission of ANY encoded signalling such as Analog modems, ISDN and ADSL, at
a concentration point such as at cable joints or an Exchange. Even a cabling
junction point in the street can be a nightmare for the Telco, a good Telco
will limit the number of ADSL/ISDN or any other form of High frequency streams,
meeting at such a cable concentration point.

As the frequencies used go up, they get hard and harder to eliminate from NEXT
situations. The exact same issue came into play when we moved from 1200bps FSK
(Frequency Shift Keying) modems, to 2400bps PSK (Phase Shifted Keying) modems.
The 2400 modem used 2 x 1200 FSK, but phase shifted them 180 degrees to get
twice the binary data through (and also why a 2400 modem is also a 1200 BAUD
modem, BAUD is an expression that relates more to the encoding format, not the
data throughput. A 9600bps modem is really 8 x 1200 BAUD modems.......;-))
Thats why latency through a 2400 modem is slightly greater than a 1200 modem,
(and greater again through a 9600 modem) it has to wait for all the phases to
be received before it can combine them back together again. This is of course
offset by the much higher binary data throughput.

9600 (or V.32 specifically) modems use 8 phase schemes, again, all at 1200bps
each. As the number of phases increased, it became harder and harder to
discriminate between the phases, the phase angle reduced from 180 degrees to 45
degrees. It required a massive engineering effort to get a receiver to work at
28k8, and do it reliably on the existing copper cables. A 56K modem actually
does a lot of this digitally now as they figured out how to get around the
issues, but they are still limited to 56Kbps as an encoding scheme total to
"plumb" the data stream through the voice part of the network.


 > > 3:  If the NT must be installed someware downstaires, you
 > > need 4 lines instad of two and that might indeed mean quite
 > > some wiring thru hidden ducts...

 >  MvdV> I can't think of any reason at all why the NT /must/

 >  MvdV> be istalled downstairs.

 > The main reason I can see here is that perhaps there is a
 > recognised demarcation point between the service provider
 > and building owner, and that is located "downstairs".

 MvdV> That sounds like on of the typical excuses our 
 MvdV> telecom used to come up with when the still had the 
 MvdV> monopoly and they did not want to tell you the real 
 MvdV> reason why they did not want to provide a service.

Yes, its still used as a reason today, however its much less an issue than it
used to be, especially if you are a large organisation spending Millions of $
per year with your providers...


 MvdV> If I understand Carol correctly in Japan telephony is 
 MvdV> still a monopoly. So one can still expect to be told 
 MvdV> "stories" there...

Sometimes the stories may be very true, but from a highly technical perspetive
and so are not always easy to explain. More often than not it all comes down to
the economic return however...

Cheers...........pk.


--- Maximus/2 3.01
 * Origin: === Maxie BBS.  Ak, NZ +64 9 444-0989 === (3:772/1)