In article <telecom24.278.13@telecom-digest.org>,
Choreboy <choreboyREMOVE@localnet.com> wrote:
> Robert Bonomi wrote:
>> In article <telecom24.276.9@telecom-digest.org>,
>> Choreboy <choreboyREMOVE@localnet.com> wrote:
>>> A relative has a farm. His phone service comes in on 700 yards of
>>> ordinary telephone cable buried along his driveway. Last week he got
>>> Bellsouth DSL. It comes in on the same conductors as before, but I've
>>> seen speeds fifty times faster than dialup.
>>> I thought 56K was the fastest speed possible with conventional
>>> telephone cable. How can DSL be so much faster with the same old
>>> cable?
>> You thought 'wrongly'. <grin>
>> "56k" is the theoretical maximum speed you can get across a (mostly
>> analog) POTS service circuit. The limit is not in the wiring, per se,
>> but in the _equipment_ (the 'switch' in the telco 'central office')
>> that that signal has to pass through. "Voice" calls, including data
>> modem, and fax, over POTS PSTN, leave your house as analog signals. at
>> the telco, the first thing that happens is that they are converted to
>> a _digital_ data-stream. this conversion is done at a rate of 8000
>> samples/second., with 8-bits of data 'precision' for each sample.
>> This means that there is 64,000 bits/second of digital data flowing
>> through the switch for a voice line. You cannot send more data than
>> that via _analog_ origin signalling, And, to achieve that 64,000
>> bit/second, your signalling must exactly match (and be synchronized
>> with) the intervals used by the analog-to-digital conversion gear in
>> the C.O. If there is _precisely_one_ analog/digital conversion in the
>> path, then, with some fancy games on the 'digital' end, you can come
>> 'close' to that 64,000 bit/sec limit, _without_ requiring the exact
>> synchronization.
> If 56K is the theoretical limit usually given, does this mean only
> seven bits are useful to the customer? It seems as if modems
> negotiate speeds in increments of 4% or so. I wonder why that is.
The answers to *that* are long, involved, and _very_ technical.
A vastly over-simplified explanation is that an A/D conversion has a
theoretical best accuracy of +/- 1/2 bit. meaning that there is an
inherent error range of a full one-bit worth
>> The _wire_, itself, is capable of passing a much broader range of
>> signals. *If* the signal doesn't have to go through the 'voice'
>> switching equipment, you are not restricted by the limits of _that_
>> equipment.
> On dialup, the farm couldn't negotiate modem speeds quite as fast as I
> could in town. I assumed the limitation was in the wire.
It is, and it isn't.
How far can you yell, and be heard? How far can you talk on a radio,
and be heard?
The technology used makes a big difference. <grin>
> That's why I was amazed to see that DSL seems to use the wire in the
> same way as dialup. Was I wrong to think the reason dialup data
> rates were slower at the farm was that the wire to the CO is longer?
"Sort-of". It wasn't the actual _length_ of the wire. If you build
the wire differently, you will get different effects over the same
length of wire.
One of the things that goes on, when you run a _pair_ of wires for any
distance is that there is a capacitance between the two wires. this is
proportional to the length of the run, inversely proportional to the
distance between the wires, the thickness of the insulation, and a
number of other factors. Capacitance between the signal-carrying
leads has the effect of 'blurring' the signal. which puts a limit on
how fast you can "effectively" change signal levels, to pass
information from one end to the other.
The _methodology_ used to accomplish the signalling determines where,
and to what extent, that 'blurring' degrades the transmission.
DSL _does_ suffer degradation with distance, just as analog POTS
modems do. If you have 15-18,000 feet of wire between you and the
C.O., you will get much slower maximum rates than if you are less than
a thousand feet from the C.O.
DSL uses a differnt _kind_ of signalling -- which calls for
differently designed transmitters and receivers -- to get the higher
data rate on the bare wires.
>> This is how DSL works, it bypasses the _voice_ switching gear. It
>> uses just the 'bare wire' between the telco C.O. and the customer
>> premises. The special eqipment in the C.O. puts a *different*kind*
>> of signal on the wires, that the "DSL modem" at the customer
>> premises understands, and the 'modem' at the customer location does
>> 'something similar', to communicate back to that special equimpent
>> at the Telco offices.
>> Voila! the limitations/restrictions of the telco voice_ switching
>> equipment are bypassed, and thus 'not relevant' to this
>> communication.
> What's the downside for the telco? With the right pricing, I think
> they could tap a huge market for increased bandwidth.
Connected to _what_? "Multi-megabit bandwidth" to the C.O. is
-useless-, unless there is "something interesting" to connect to.
Who needs the capability for a dozen or two (or more) simultaneous
voice telephone calls from their house? For anything other than
voice, you have to have that 'something else' available for access at
the point that that high-capacity circuit from the customer premises
terminates.
Getting _to_ the C.O. from the customer premises is the 'cheap' part.
Whether it is POTS, or DSL, or whatever. Amortizing the equipment
over, say, 5 years, you're talking about circa $3-5 dollars/month.
The 'wire' cost, amortized over the useful life (50 years+) of the
wire pair, is of the same order, maybe a bit lower.
All the rest of the money goes towards "what to do with it, _after_ it
arrives at the C.O." For voice, to calls to a number in a different
C.O, you have to have inter-connects to get the call _to_ that C.O.
If there isn't a direct trunk circuit, you have to go through another
layer of switches (called "tandems") -- at least one, possibly several
-- to get to the destination 'local' switch. 7 figure price-tags, per
unit.
Same thing connecting to the 'internet' -- you generally have to 'pay
somebody' to pass your traffic on to the rest of the world, Costs for
that depend on "how much" traffic you have. More traffic, more
cost. and bigger, more expensive equipment.
Robert Bonomi (bonomi@host122.r-bonomi.com)