V.90: mu-law vs. a-law?

[David C.]

[email]geos@epost.de[/email] (Georg Schwarz) writes:

>
> Does V.90 require the explicite configuration by the user of mu-law
> or a-law, respectively? I'm ujsing an iBook's internal modem in
> Brazil with a German version of Mac OS X. So far I only get V.34
> connects, not V.90 (could be the line quality though).

It's line quality.

mu-law vs. a-law is a function of what the phone company uses on the
digital trunks. Your modem connects to an analog circuit and so
isn't involved in this.

If you are running the server side of a 56K connection (which is
attached to a digital telco circuit) then it would make a difference.
But the result of using the wrong encoding would be no connection at
all, not a reduced-speed connection.

> Does anybody know which would be the appropriate V.90 setup for
> Brazil?

Can't help here. I don't know what Brazil's telco standard is.

-- David

[Georg Schwarz]

David C. <shamino@techie.com> wrote:

> mu-law vs. a-law is a function of what the phone company uses on the
> digital trunks. Your modem connects to an analog circuit and so
> isn't involved in this.

I'm not so sure about that last sentence. V.90 in fact does make use of
the fact that the subscriber is connected to a digital switch, and it
has to make use of the digital-analog conversion internals. Otherwise
with the bandwidth of an analog line we would be limited to V.34. For a
strictly analog connection with the voice quality bandwidth (spectrum)
available, 33600 bps basically is almost the theoretical limit.
At least Apple's V.90 modem (and I'm sure many others as well) has some
settings of assuming either a-law or mu-law analog/digital conversion
facilities. What I'm wondering is whether this has to be set manually or
whether it is detected automatically by the V.90 protocol.

>
> If you are running the server side of a 56K connection (which is
> attached to a digital telco circuit) then it would make a difference.

normal V.90 modems cannot even run as server side modems (if you try
V.34 is the most you get). On the other hand, you are right, for V.90
server side equipment it surely does make a difference.

> But the result of using the wrong encoding would be no connection at
> all, not a reduced-speed connection.

I assume if I don't get a V.90 connection the modem falls back to V.34.
This is what I get.


--
Georg Schwarz [url]http://home.pages.de/~schwarz/[/url]
[email]geos@epost.de[/email] +49 177 8811442

[David C.]

Georg Schwarz writes:

> David C. wrote:

>>
>> mu-law vs. a-law is a function of what the phone company uses on
>> the digital trunks. Your modem connects to an analog circuit and
>> so isn't involved in this.

>
> I'm not so sure about that last sentence. V.90 in fact does make use
> of the fact that the subscriber is connected to a digital switch,
> and it has to make use of the digital-analog conversion internals.

Not correct.

It assumes that the server side is digitally-connected to the public
phone network.

The client side is always connected to an analog loop. If you have a
digital connection in your home, then you have ISDN or a leased line,
and your interface is nothing like a modem.

It is true that your local line must connect directly to an
analog-digital converter, but that's because analog repeaters
introduce too much noise into the signal. It is not a function of
the encoding algorithms.

>> If you are running the server side of a 56K connection (which is
>> attached to a digital telco circuit) then it would make a
>> difference.

>
> normal V.90 modems cannot even run as server side modems (if you try
> V.34 is the most you get).

Didn't say otherwise. When you have a digital connection to the
phone system (as in ISDN, leased lines or other digital connections)
then you must generate the bit-stream that corresponds to the audio
that an analog client will get. In order to properly do this, you
absolutely need to know if it's a-law or mu-law.

This is the case for any protocol - even the old Bell-212A (300bps)
protocol - if you have a digital link to the phone network.

a-law and mu-law are analog-digital encoding standards. Equipment
which converts between the two must support the right standard.
Which means modems that have digital connections to the phone network
and the equipment that telcos use to terminate the analog local loop.

>> But the result of using the wrong encoding would be no connection
>> at all, not a reduced-speed connection.

>
> I assume if I don't get a V.90 connection the modem falls back to
> V.34. This is what I get.

Most (if not all) modems manufactured have firmware that supports all
prior standards (33.6, 28.8, 14.4, 9600, 2400, 1200 and 300). They
negotiate a protocol at startup using the answer tones and the
high-speed protocols have facilities for dynamically changing speeds
during a session in order to accommodate line-condition changes.

But none of the standards (not even 300bps) will work if a digitally-
attached modem uses the wrong encoding standard.

-- David