From: floyd@tanana.polarnet.com (Floyd Davidson)
Newsgroups: comp.dcom.telecom.tech
Subject: Re: Details on B8ZS line code used in T1
Date: 18 Apr 1998 01:40:50 GMT

Kelly Christian Swanson  <kswanso@uswest.com> wrote:
>when i said "T1 repeater", I pretty much meant any device that would generate a
>T1 signal, be it the ds1 side of a fujitsu multiplexor, an HDSL unit in the
   [snip]
>opposed to classic repeaters (although they are still used a lot, see below).

A DS-1 interface!  (Which normally should really be distinguished from a
T1 repeater, but for the purposes of this discussion it doesn't make any
real difference.  Both must do the same thing in regard to B8ZS and timing.)

>... needless to say, you often have data and voice channels
>coexisting on the same T1, so ones density is an issue we
>always have
...
>As far as many many miles out of a CO... you generally need a
>T1 repeater every 6000 feet for a metallic 4 wire pair, and
>that somewhat depends on gauge and characteristics of the pair.

For the original T1 carrier span line, the first repeater is
3000 feet from the CO, the rest are 6000 feet apart, and the
tail should be less than 3000 feet if possible.

However, only the DS-1 interface side of a T1 facility is
standardized, and the actual span line equipment can be and do
just about anything on the cable pair.  A number of companies
now sell equipment which can do such things as provide a "T1" on
a single pair or handle a few miles of cable without a repeater.
Of course none of these equipment types can interoperate with
those made by different manufacturers, while there are a
relatively large number of companies that make equipment
directly compatible with traditional T1 span line equipment.

>> When is a T1 repeater used (anymore)? I thought these got rolled up to
>> DS3, then to optic fiber and rolled back down. I suppose some customer's
>> T1s require repeaters, but isn't this rather arcane in the 1990's?

Fiber will not totally replace copper cable for many decades.
It is far cheaper to install T1 facilities on existing copper
cable to increase circuit density than it is to install a fiber.
That is most particularly true when the number of DS1's would be
less than a single DS3 with 28 each DS1's.

>> I have often pondered why there is such an aversion to 15 consecutive 0s
>> and why ZCS and B8ZS are prescribed/configured for so-called voicegrade
>> T1s in the US, which use mu-law companding that cannot result in a PCM
>> value of all zeroes. Aren't most T1s good for many, many miles out of
>> the switch, and only require repeaters when the T1 hops off copper/fiber
>> onto airwaves?

While there are, as I mentioned above, T1 span line units that
do work out to several miles without repeaters, most telephone
companies are pretty much staying with the traditional, and
interchangeable, span line facility designs.  If nothing else
that reduces the amount of expensive test equipment necessary to
maintain the installed facilities.

Since it is common practice to intermix non PCM voice grade
circuits on T1 facilities, the common practice today is that all
new facilities are configured for B8ZS unless there is a specific
reason to use AMI (such as interfacing to old equipment which
cannot be configured for B8ZS).

The ones density requirement has a hidden benefit which is not
widely known.  The requirement exists because a DS1 receiver
clocks data into a buffer using a clock derived from the input
signal.  As Kelly's diagram showed so well, all 1 bits are
pulses, and 0 bits are no voltage, and hence some ones on the
input signal are required for the phase locked loop circuit to
recover timing well enough to generate a clock signal.  But
those are just the dirty details, and the benefit comes because
the buffer is dumped out using an entirely different clock,
which is external and is _frequency_ locked to the world wide
network.

Because the data is buffered, those two clocks do NOT need to be
phase locked, and only need to be the same frequency.  The two
can vary in phase enough to wander around much more than the
time interval for a single bit.  If timing were not recovered
from the input signal, then the phase of the external timing
source would have to be within less than 1/2 a bit interval or
errors would occur!  As is, the phase wander can be 1/2 of
_buffer_ interval (commonly a 2 frame buffer size is used)!

Obviously things like transcontinental T1 facilities would be
impossible, and T1's over satellite systems wouldn't come even
close if a phase locked external reference were required.  So
the 1's density requirement is essentially what makes digital
carrier systems as useful as they are.

  Floyd

--
Floyd L. Davidson                                floyd@polarnet.com
Ukpeagvik (Barrow, Alaska)