RE: [802.3ae_Serial] WIS TJP Document revision
Tim,
A number of comments.
1) These should not be labeled "Jitter Test Patterns" since they are to test
a variety of parameters. The square wave pattern isn't used for jitter test.
The title should be "Test Patterns." This affects text beyond the title .
2) The receiver should be required to receive and report errors in the
framed mixed frequency pattern. This is necessary in order to run tests on
the receiver's tolerance of jitter. There should be no requirement and no
recommendation that the receiver be able to receive the square wave pattern
as it is used only for testing transmitter characteristics.
3) The annex should not state requirements on which signal is used for which
test. Clause 52 will state the required pattern when it specifies the
parameter. The annex might mention the uses of the tests in an informative
manner such as "The square wave test is used for transmitter tests which are
best done with a repeating waveform such as rise/fall time and extinction
ratio. The framed mixed frequency pattern test is used for transmitter and
receiver tests which require stresses such as long run length and varied
frequency content such as jitter tests." Rather than a long list of cross
references to clause 52 subclauses, a general statement such as "See clause
52 for specification of the appropriate test pattern for each parameter."
4) The meaning of: "The PRBS section shall start at a reset TBD value at
the location immediately following fixed stuff in the path overhead of the
first frame." is not entirely clear to me. Does it mean that the PRBS
generator is seeded with the value or is it the value of the first n bits
out of the scrambler. I think the former is better to use but it needs to be
stated.
5) A value for the J1 byte needs to be specified. The meaning of "optimal
stress" is not clear. Does that mean worst case stress such as 8 more bits
matching the CID? One might make J1 0xAA or 0x55 which would introduce an
abrupt frequency content shift.
6) Shouldn't J0 be specified to have a fixed value for this test since it
falls between the A2s and the CID? What about the other octets that say "per
T1.416" ? If one is going to be downloading a pattern into the tester, we
can't let their values fluctuate as they may when reporting error rates,
etc.
7) It is extremely unlikely that any of the components which are affected
by running disparity have a memory as long as a frame so neutral disparity
across a pair of SONET frames is not important. Also note that the frame has
150,336 octets of PRBS which are further scrambled by the SONET
scrambler and should have close to neutral disparity
384 octets of A1/A2 which together have neutral disparity
182 octets of Z0 with neutral disparity
9 octets of CID with all 1's or all 0's disparity
4617 other overhead octets which are scrambled by the SONET scrambler
and should therefore have close to neutral disparity.
The 9 octets of CID produce less than 0.01% of imbalance across a frame so
even if the components do have a long memory for disparity compared to a
frame, this isn't enough imbalance to be significant.
Pat
-----Original Message-----
From: Tim Warland [mailto:twarland@nortelnetworks.com]
Sent: Monday, June 25, 2001 1:43 PM
To: serialpmd
Subject: [802.3ae_Serial] WIS TJP Document revision
Prior to tomorrow's conference call, I wanted to circulate
the revised WAN jitter test plan.
It is getting close to decision time. The attached document
is one possible test method. Draft 3.2 of the standard also
has a WAN jitter test plan. Finally we have a question
on whether people would use SONET based testers or
bit based testers.
I am neither a system integrator nor an optical engineer.
The WAN JTP is lacking participation from people with
experience in these areas. We have two more conference
calls prior to Portland to develop a solution.
Looking for your feedback.
--
Tim Warland P.Eng.
Hardware Design Engineer Broadband Products
High Performance Optical Component Solutions
Nortel Networks (613)765-6634