Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

RE: Question on XSBI




Pat,

You're right - even if the TX side of the PMA doesn't implement the
full frame synchronization process that the RX side must perform, it
still needs to look for the A1/A2 boundary. Thanks for the clarification.

Best regards,

- Tom

-----Original Message-----
From: pat_thaler@agilent.com [mailto:pat_thaler@agilent.com]
Sent: Monday, February 26, 2001 4:28 PM
To: gnicholl@cisco.com; Tom Alexander; stds-802-3-hssg@ieee.org
Subject: RE: Question on XSBI 

Tom,

I'm afraid Gary isn't quite correct about the transmit side. Though the XSBI
signals on transmit are byte aligned, the LW-4 PMA has to find the Sonet
frame A1/A2 alignment (see page 458 around line 43 and figure 53-5 on page
467). It is important that the first bytes of A2 go to the Lane 0. If the A2
could start on any lane, the receiving LW-4 PMA at the other end couldn't
sort out the lanes to remux the 4 signals into a single data stream
accurately. The WIS output is byte aligned on the transmit side, so the
transmit sync machine only has to search for A1 and A2 with a fixed byte
alignment, but it must do the search. That is why the transit sync state
machine doesn't have the hunt state to do byte alignment and goes straight
into finding the A1/A2 transition.

The LW-4 PMA does the A1/A2 alignment and on the receive side needs to do
the whole bit and byte alignment. Therefore, no additional layer is required
to provide framing.

Pat

-----Original Message-----
From: Gary Nicholl [mailto:gnicholl@cisco.com]
Sent: Thursday, February 22, 2001 2:42 PM
To: Tom Alexander; stds-802-3-hssg@ieee.org
Subject: RE: Question on XSBI



Tom,

That is what I suspected. I agree that the A1/A2 alignment is probably
automatic coming out of the WIS. I was actually more concerned about what
happens in the receive direction of an ELTE transponder, where the output
of the line side 16:1 demux would be fed into the LW-4 PMA. The output of
the 16:1 demux is not A1/A2 aligned. I guess a additional framing layer
would be required between the demux and the LW-4 PMA in order to perform
the necessary A1/A2 alignment ?
Or maybe no one is considering developing an ELTE with an LW-4 interface ?

Gary ....


At 05:19 PM 2/22/01, Tom Alexander wrote:
>Hi, Gary,
>
>In the transmit case, the WIS will send data on the XSBI with the proper
>A1/A2 alignment such that the LW-4 PMA will not need to do framing. This
>is actually automatic (you have to go to some trouble to NOT do this!),
>but Clause 50 specifies this anyway for completeness. In the receive case,
>the the XSBI receive path into the WIS does not require any A1/A2
>alignment; however, the LW-4 PMA - but not the serial PMA - will have to
>do basic A1/A2 framing so that it can align the incoming streams.
>
>Best regards,
>
>- Tom Alexander
>
>-----Original Message-----
>From: Gary Nicholl [mailto:gnicholl@cisco.com]
>Sent: Thursday, February 22, 2001 12:52 PM
>To: stds-802-3-hssg@ieee.org
>Subject: Question on XSBI
>
>
>I have been reading through clause 53 and it triggered a question on XSBI.
>
>Are the 16 data bits of the XSBI interface in any way aligned with the byte
>boundaries of the SONET frame ? I don't believe that any alignment is
>specified in the OIF SFI-4 spec.  If this is the case does it mean that the
>LW4-PMA has the ability to locate the SONET A1/A2 frame boundary on any one
>of 16 possible bit alignments within the XSBI interface ?
>
>Gary .....