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

[802.3ae_Serial] RE: SJTP: WIS Pointer Processor




I agree with Tim. Just about every SONET person I've talked to
on this subject expressed surprise at the notion of trying to
sync to an arbitrary pattern. The industry standard apparently
is to include SONET framing within the BERT, such that the
payload could be properly extracted and checked. However, when
I presented the WIS jitter test pattern proposals in St. Louis,
a strong message was received from various people to the effect
that they wanted to program up a bit sequence into the BERT
and sync on that.

We should probably put this to bed first before continuing on
the WIS jitter pattern work. If the conclusion is that the WAN
and LAN PHYs will take different approaches (not just different
patterns) to the test equipment, then there's no need to invent
all these patterns for the WIS - just point at the ITU docs that
Tim has thoughtfully provided, without modification.

- Tom

-----Original Message-----
From: Tim Warland [mailto:twarland@nortelnetworks.com]
Sent: Wednesday, June 13, 2001 8:46 AM
To: Ben Brown
Cc: stds-802-3-hssg-serialpmd@ieee.org
Subject: Re: SJTP: WIS Pointer Processor



Ben Brown wrote:

<snip>

> For the WAN pattern, this is not the case. Your proposal already
> suggests we're using 29% of the total PRBS but we're complicating
> the process by starting the PRBS at a defined location on every
> other SONET frame.
>
> I'm being told by AMCC internal sources that BERTs exist which
> are able to synchronize to the SONET frame then synchronize to
> a PRBS within the payload area of that frame. If this is the
> case, why can't we simply let the PRBS run from any known
> starting location? It would simplify the generation of the
> payload in regards to timing it to the SONET frame. It would
> also no longer require ignoring a group of errors at the start
> of every other frame.

This is correct (again). There are SONET based BERT testers
which remove the payload from the overhead, then perform
synchronization and report BER. For these machines, and for
the WIS receiver for that matter, a free running PRBS generator
works well.

The discussions we had within the WIS jitter ad hoc focused
on BERT test equipments which were not SONET based. I refer
to these as bit based BERTS. A bit based BERT has no concept
of SONET frames. It examines each bit individually, so it must
know what each bit is going to be. If the PRBS is allowed to float
within the SONET frame, the bit based BERT can not be
expected to know this relationship. It would report the expected pattern

 not matching the bit sequence with which it was programmed.
This was my justification for "locking" the PRBS within the
SPE - in which case every bit within the two frames can
be predicted in advance.

It is up to the SJTP ad hoc to define the test pattern and therefore
the type of test equipment which would be required. Should we
require the use of SONET based BERTs for 10GBASE-W, or
require a bit based BERT for 10GBASE-R and 10GBASE-W?
As Ben says pattern generation and checking is easier with the
SONET BERT.

--
Tim Warland     P.Eng.
Hardware Design Engineer  Broadband Products
High Performance Optical Component Solutions
Nortel Networks                (613)765-6634