Re: 56k Required encoding

Michael Brennen (mbrennen@fni.com)
Sat, 26 Jul 1997 08:36:36 -0500 (CDT)

My understanding is that AMI was and is used for voice T1s, and B8ZS was
brought about by the need for data spans with long patterns of null data.
This is from experience in the telephony industry; my involvement with
this was back in the mid 80s, so my memory could be fuzzy. Your statement
"with the proper one-bit density" seems to support this.

My T1 span was incorrectly configured to AMI/SF when installed; I tried to
use it by configuring my end equipment to AMI, but it would not run error
free; high traffic rates (ping floods with differing data patterns)
greatly reduced the errors but never eliminated them. B8ZS/ESF cleaned it
up with no other mods.

-- Michael

On Sat, 26 Jul 1997, Dan Swartzendruber wrote:

> At 11:52 PM 7/25/97 -0500, you wrote:
> >
> >AMI is not a good choice for a data circuit, and I'm not sure the PM3 even
> >supports it. It has been some time since I was around the details of
> >this, but as I remember it, AMI does not have enough line transitions to
> >allow proper clock recovery. Unless the line is very busy (to keep
> >transitions high) you will get errors on the link. B8ZS forces a high
> >enough line transitions rate through enforced bipolar violations to allow
> >decent clock recovery, and should always be used on a data circuit.
>
> This is incorrect. B8ZS will get you the maximum 1.544 mb/s data rate,
> but AMI is perfectly usable (how on Earth do you think people used T1
> circuits before B8ZS was available?) Using AMI, you can set each DS0
> to be 56K instead of 64K. This gives you an aggregate 1.3 (approx) mb/s
> data rate, with the proper one-bit density. Any decent T1 CSU/DSU will
> allow you to configure this.