I have just the opposite problem :)
Here is a repost of a previous thing I put together.
A T-1 is a T-1 is a T-1.
Hopefully I can help clear up some confusion. I am going to talk
about framing, timing and signaling.
Framing
A T-1 frame consists of 193 bits (24 timeslots * 8 bits + 1 control
bit). Each of the 24 inputs is assigned a fixed time slot (this is
synchronous time division multiplexing). Framing bits are used to
create a pattern to help synchronize the equipment. The pattern of
framing bits in successive frames serves to define superframes. Two
superframe formats are used: D4 (SF or superframe) and ESF (Extended
SuperFrame).
In SF, the framing bits of the odd numbered frames are used to
create a 101010 pattern for the receiving end to maintain
synchronization. The framing bits in the even numbered frames are
used to create a pattern of 001110 for the receiving end to know
which frames contain the signaling information for the T-1. The
eigth bit in the 6th and 12th frames are robbed to provide signaling
for the timeslots. SF is mainly used in voice.
In ESF, only 6 of the 24 framing bits are used to maintain
synchronization, leaving 18 bits to be used for other purposes.
Specifically, 6 of the 18 bits are used for error detection (CRC),
and the remaining 12 bits are used for data link maintenance or
diagnostic information. Signaling information is robbed from the
6th, 12th, 18th and 24th frame for voice channels only. The only
exception is if signaling is done out of band (ie D channel in ISDN,
or SS7).
Timing
There are two issues in timing. First, we need a way to provide the
receiver with the clock it needs without a separate timing signal.
The approach is to use a line signal format that permits the
receiver to synchronize its own transmit clock from the one bits in the
signal stream. This, in turn, dictates sufficient "timing events" in
the signal - a requirement that translates into the need for
adequate "one's density" (this is covered in signaling).
Second, we need a way to maintain synchronization throughout the
network. This requires clock standards and agreements on timing
relationships between various network elements.
Signaling
T-1's use AMI (Alternate Mark Inversion), also known as BPRZ
(Bipolar Return to Zero). In AMI, zeros are well represented at zero
volts. Ones are represented with alternating pulses, each pulse
occupying 50% of the bit time (duty cycle). Occasionally, the
receiver will detect two consecutive ones without an alteration of
the voltage in a row. This is a bipolar violation.
Devices that receive T-1 signals derive timing from the line
signal. Using AMI means that zero bits are "flat" and therefore
provide no timing information. There must be enough ones in signal
to maintain synchronization. Most equipment is designed to maintain
synchronization through 15 consecutive zeros. This only becomes a
problem with digital data as the MU-255 PCM encoding does not create
any all zero words.
There are three ways to ensure ones density. First, there is MU-255
PCM encoding (used in voice).
The second is ZCS (Zero Code Suppression). In ZCS, the least
significant (eighth) bit of every time slot is set to a 1 (this only
allows 56k for data).
The third is B8ZS (Bipolar with 8 Zero Substitution). B8ZS allows
all 64k of a timeslot to be used by data without regard to the
number of zeros in the data. This is acheived by substituting a
special pattern for an all zero word. The special pattern is
00011011. How can the recevier determine whether the transmitter
actually sent 00011011 or the special B8ZS pattern representing
00000000? The solution is to use deliberate violations in the
alternating polarity scheme to mark the B8ZS pattern. These bipolar
violations occur in bits 4 and 7 of the B8ZS word.
In conclusion, a T-1 is just a physical transport. The real
difference is in the services provided by T-1's and the equipment
attached.
When I hear Channelized T-1, all this means to me is that the
equipment on the line possesses the capability to treat each time slot
as an individual channel (whether you run voice, or mux 24 56k/64k
circuits onto it).
PRI is 23B+D ISDN delivered via T-1. One end of the circuit
terminates into a switch that contains the line cards and software
necessary to switch the B channels and deliver control information
over the D channel. Each channel is just an individual time slot on
the physical transport - T-1.
I know this is rather long, but I hope this will clear some things
up for people.
Kyle Platts
CSS-Tech
!NTERPRISE Networking Services
Footnote: I used the DataComm/2000 book from Hill Associates as
a reference.