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

RE: Clause 46 - Preamble




Thanks Pat, it wasn't my intention to imply anything on coding limitations
of either of the current PCS.  I should have waited for Shimon's reply on
the topic, he states the rationale much better than I did below.

--Bob

-----Original Message-----
From: THALER,PAT (A-Roseville,ex1) [mailto:pat_thaler@agilent.com]
Sent: Friday, February 16, 2001 12:48 PM
To: Grow, Bob; 'Gareth Edwards'; HSSG
Subject: RE: Clause 46 - Preamble


Bob,

To clarify, there is nothing in either PCS that prevents encoding both Start
and SFD in the same column. Preamble and SFD octets are just data characters
to them and they do not examine the content. So, it would only be some
hypothetical future PCS that might not encode Start and SFD in the same
column.

Regards,
Pat 

-----Original Message-----
From: Grow, Bob [mailto:bob.grow@intel.com]
Sent: Friday, February 16, 2001 10:37 AM
To: 'Gareth Edwards'; HSSG
Subject: RE: Clause 46 - Preamble



Gareth:

I think D2.1 faithfully implements the Response to comment #252.  The
response is an AIP for two main reasons, we didn't accept Shimon's proposed
text, and the response is to not process the frame, where Shimon's suggested
remedy was to cause a CRC error.  

It wasn't felt to be any more difficult for the RS to recognize the SFD in
the same column as the Start.  If some PCS types are not able to encode both
Start and SFD in the same column, the RS doesn't need to further enforce it.
There is nothing that a conforming implementation with the current PHY types
will do to change the preamble length, but why add an unnecessary
restriction to eliminate future options.  

--Bob Grow

-----Original Message-----
From: Gareth Edwards [mailto:Gareth.Edwards@xilinx.com]
Sent: Friday, February 16, 2001 7:56 AM
To: HSSG
Subject: Clause 46 - Preamble



Afternoon all,

Shimon raised a comment against D2.0 SC 46.2.3.3 (#252) asking for the
removal of the requirement that the preamble be a fixed length on
receive; this was accepted in principle. However, I seem to remember in
the discussion on this comment at Irvine that there was agreement that
the Start control code and SFD couldn't appear in the same RXD column;
there would be one or more clock edges between the Start code and SFD. 

The new text (in D2.1 SC 46.3.3.3) does not explicitly rule in or out
the appearance of the SFD in the same column as the Start code; this is
consistent with Shimon's original suggested remedy.

Is this something that has slipped through the rewrite or has False
Memory Syndrome struck again?

Cheers
Gareth

--
/ /\/\ Gareth Edwards              mailto:gareth.edwards@xilinx.com
\ \  / Design Engineer
/ /  \ System Logic & Networking   Phone:   +44 131 666 2600 x234
\_\/\/ Xilinx Scotland             Fax:     +44 131 666 0222