Ascend Archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
(ASCEND) V.90 and telcos
I also sent a similar report to bungi.com, but since I am not a registered
ascend user, I think it gets dropped.
Here was what I did to fix the same problem. This is a letter to Walt
Mossberg at the Wall Street Journal.
Last January I upgraded to a 56K Flex modem using the Rockwell chip set.
The choice of a modem that uses the Rockwell chip set was based on
compatibility issues. Clark-Verio uses Ascend Max 4004 Modems which also
use the Rockwell chip set. However I could not get the modems to establish
a communications channel. The units would connect and then drop the
connection after the carrier was detected at each end. I had similar
problems even when trying to connect to the Ascend modems at 14.4k using a
Powerbook when they were installed in their present location.
Clark's solution was to append commas to the dial string which forces a
connection at the V.34 speeds, hardly what I was expecting since I paid
good money to buy a 56K modem. The interesting thing was that when I tried
connecting from a Powerbook from Kill Devil Hills, N.C.(via AT&T toll
circuits) It always worked even to the Ascend modems!
After I got the 56K modem and had problems I tried calling a different
number that Clark had for accessing 56K modems, but one that required going
through the AT&T toll circuits. Every connection was completed at speeds
greater than 40000 BPS. I contacted Bell Atlantic and after quite a bit
finally found some one who was able to at least shed light on the problem.
(Bell Atlantic insisted on checking the lines which they did several times
before it dawned on them that I use the same wires to the telephone company
for local as well as toll calls!)
My local switch for the 997 exchange is a Northern Telcom DMS100 which
digitallly connected to a AT&T ( Now Lucent) 5ess switch to which the
Ascend Max Modems were connected.
The problem simply stated is that the signal compression algorithmms used
by the DMS100 changed the signal characteristics of the signal sufficiently
that the connection would be dropped soon after the carrier was detected by
both modems.
The digital pad used by the AT&T toll circuits identified these "Digital
Impairements" and corrected them. Hence the ability to connect via the toll
circuits.
I contacted Dean Grumlose at Rockwell with an engineer Norman Apati there
conducted some tests using their new V.90 code which I loaded into my
modem.
Testing was not straight forward, because Ascend chose to install the V.90
code from Rockwell at the same time they made a version change to the code
in their Modems. The version change was far from robust and many ISP's
refused to upgrade to the V.90 code. (At this time S&P was adding Ascend to
the S&P 500!) However the V.90 code in my modem did correctly identify that
the ISP uas not using V.90 and it correctly defaulted to 28.8K.
Rockwell's V.90 code has a better signal prossessing algorithm that
corrects for this problem. Since I have been using V.90 ( I had to bully
Clark to install it since they were understandably reluctant to install
6.0) I connect reliably at the V.90 speeds. Version 6.1.3 of the Ascend
code seems to work fine at least from my end. I love connecting at the
higher speeds.
Since the W98 internet code seems to be better at reporting the DCE speed
rather than the DTE speed, I think there are going to be many users that
are going to complain when they they find out they are really connecting at
28.8K rather than at the faster speeds. When I first reported the problems
to Clark back in September, They suggested using commas to force the slow
connect and set my machine to report the DTE speed rather than the DCE
speed.
By the way the V.90 speeds are from 28.8 K to 56 K in increments of 8000/6.
The way the modem scripts work for Open Transport, the report from the
modem is "matched" with speeds in the script. Without a revision to the
script, the usual report is "Communicating at an unknown rate." (Using a
terminal window when connecting that does not use the script will verify to
the user that the speeds arre actually V.90 speeds.)
If you have any questions email me or call 410-997-9595.
Robert Weeks
++ Ascend Users Mailing List ++
To unsubscribe: send unsubscribe to ascend-users-request@bungi.com
To get FAQ'd: <http://www.nealis.net/ascend/faq>