Re: K56Flex vs. X2 (Boardwatch article)

Michael Whisenant ( )
Thu, 26 Mar 1998 18:15:50 -0600


Without disclosing anything that is currently covered under an NDA, here
goes some thoughts.

If the problem is in the local loop, how can one bypass the local loop?
For example have you had customers that can call USRs toll free testing
number, find that their lines are x2 (which also means K56Flex) capable,
yet when they call your local number never seem to get a 56K connection?
If you were to monitor their line you would see a much lower db of loss on
the local loop vs the long distance loop. If you have more than 1
switch/wiring center in your town and you had a number in the other wiring
center that terminated in the original wiring center, it just might have a
different loss rating. The call path would be from the users demarc to
his/her wiring center to the second wiring center back to the original
wiring center.

This problem could also be a result of the interoffice trunking. We have
been on occasion able to monitor the exact call path. It seems that if in
the interoffice trunking that the call leave a co on B8ZS and then passes
across AMI and back to B8ZS (nothing you can do or control about this) that
you are more likely to have problems. I have not gathered enough data to
make certain claims about the AMI/B8ZS interoffice issues, but BellSouth
seem interested in this aspect. The key is in the local codec and line
loss. If the db of loss is less than 3bd and higher than about 10~12db you
have problems. Those wil lower loss can use the method described above.

The other thing to improve the speeds is to modify the transmitter output
db signal strength on your terminal server. It is hard to say which way to
go, and you just need to make minor adjustments and evaluate. On lines
from 1 telco I have the power set to -11db (default for USR), while on
other lines it works better at -9db to -8db. I have some in GTE country
that I have set to -13db to get the speeds up. These are just some of the
issues, I would recommend you know the S-registers for the modems dialing
into your POP and try different values there if you really want to tweak.
Since I was USR x2, I had Sportsters and Couriers, and then the generic TI
DSPs to work with. Not that bad. I am having more troubles with the
Motorola v.90 modems that I ever did with x2 models. Considering Motorola,
(up to a month or so ago) built the majority of their modems here I still
can not get an engineer to address this issue.

>Michael Whisenant <> writes:
>> Actually the real issue is in the local loop. We have worked with
>> BellSouth, Seimens, Nortel, Lucent, USR and found several keys in how to
>> improve the chances of reliable 56K connections. Like if the subscriber
>> line is build out at less than 3db what steps to take, vs a customer site
>> that is over 10db loss.

>> We have challenged any local, regional, or national ISP to
>> outperform our connect and throughput speeds and no one took our
>> challenge. The regional or national could not really afford at
>> 19.95 to match our level of service, and the other locals decided
>> not.
>Maybe you could share some of your discoveries? This list is for
>people with similar problems and neesd to communicate and help each
>other, not just advertise our own superiority.
> ----------------------------------------------------------
> NTISP Mailing List