[802.3ae_Serial] From Serial PMD 24 July: revised issue list; strategy of stressed sensitivity needs agreeing; jitter; ...
We added a few items to the issue list, and discussed mainly jitter and
strategy of stressed sensitivity.
Raised significance
-------------------
Jitter specs
SONET receivers said not to work with IEEE draft jitter specs, and a
belief that the SONET specs aren't complete: worst legal Tx and Rx might not
make a working link. Suggested change of ER/EW DJ from 0.35 to 0.3 UI, on
the basis that the optical receiver (TP3 to TP4) needs 0.1 UI of DJ, and
that the ER/EW transmitter would use a modulator type transmitter rather
than a directly modulated laser. Both these points were debated: optical
receivers may add significantly less than 0.1 UI of DJ, and it would be a
pity to block out future directly modulated transmitters at 1550 nm. Expect
the feasibility ad hocs to bring jitter numbers to the table. Individual
PMA vendors are reluctant to show their numbers in public but could somehow
go further towards saying what is industry-typical. Comment that a hard
demarcation at TP4, with measurement tolerances on each side, would cost
some DJ budget.
Question: has anyone tried the jitter measurement procedures as in the draft
standard? Some have done something similar, using coax cable and a limiting
amplifier to create DJ, and standard 8B10B and 2^31 patterns, but no-one was
willing to report that they had followed the procedure exactly, and no-one
said they had tried the new test patterns. I was asked to tell the group
where the patterns were kept.
[Here is a repeat of my email "Whereabouts of the serial test patterns" Thu
26/07/01 15:14 +0100:
John Ewen sent the patterns to this reflector at Mon, 2 Jul 2001 17:44:05
-0500, Subject: Re: [802.3ae_Serial] Re: SJTP Patterns. The email, with
attachments, is filed at
http://www.ieee802.org/3/10G_study/public/serial_adhoc/email/msg00395.html .
The x.txt attachments have been renamed as y.bin.
We are asked to try out these in pairs, e.g. A and B alternating, B and D
alternating, etc. We should compare them with standard PRBSs, traffic, and
an idling link. We should look for differences in apparent sensitivity, DJ
and RJ for different patterns/signals.]
Items added
-----------
Choose mask that would allow elimination of 850 nm rise/fall time spec
Discuss Mike Dudek's concern around 1310 mask/rise/fall
Possible adjustments to 1550 power budget: setting overall power up or down?
Tim Warland (ER/EW feasibility group) expects to report on this in a
couple of weeks.
Inconsistency between stressed and unstressed sensitivity
Unstressed sensitivity = Tx power - "budget", is set with regard to
margin for unidentified penalties, as is typical in optical link design.
However, the stressed sensitivity ignores the budget and the margin: it is
Tx power - (all?) losses and impairments. This seems confusing. The
consensus on the call was that setting both sensitivities with regard to the
same amount of margin would make more sense. But it would have to be a
reasonably small margin! Further, not all penalties should be treated the
same when calculating the stressed receiver sensitivity: some are measured,
some are not, some are "the transmitter's fault", some are "the receiver's
fault". I'll need to think about this. It's not a very big issue in dB:
the unmeasured penalties are reflection noise, modal noise and polarisation
mode dispersion.
As a group, we should get consensus on what our strategy for stressed and
unstressed sensitivity is quickly, before going back to discussing the
numbers.
Progress on issue list below
----------------------------
# 363 Making ER/EW optical specs comprehensible: it is not clear in the
draft that the penalty TPD needs to be taken into account in obtaining the
stressed receiver sensitivity. Needs a comment with remedy for D3.2.
To get these optical specs straight and consistent we will have to think
more in terms of self-consistent sets of numbers: a circulated proposal
would be nice, and comments against future drafts should offer complete sets
of numbers. I will try to issue a spreadsheet populated with the numbers
which comment resolution has mandated for draft 3.2.
# 367,368,130 LR/LW maximum average transmitter power and receiver overload:
The opposite of progress. I have finally done my homework of the linear
programming diagram of Tx setup, and I find that for legal low extinction
ratios, the setup window narrows to only 4 dB, as compared with the 5 dB
intended (and 8 dB in 1000BASE-LX).
Other
-----
Dispersion measurement? Someone recommended G.650 - Definition and test
methods for the relevant parameters of single-mode fibres. Sorry I didn't
catch what it was recommended for.
Here's the issue list repeated from my email of 23 July:
Serial PMD ad hoc issues list
-----------------------------
# numbers are from comment database
General
-------
Liaison letter from ITU-T SG15
Optical power measurement in network maintenance
(Technical feasibility - covered by other groups)
All wavelengths
---------------
# 317 combine Rx jitter tolerance and Rx sensitivity tests
further work (add calibration sequence, etc.).
Road-test jitter and stressed sensitivity measurements
Jitter spec values
Road-test the test patterns (LAN and WAN?)
SR/SW
-----
# 77
Check SR/ SW Rx sensitivity value. May need to allow for jitter penalty.
# 78
Check SR/ SW stressed Rx sensitivity value. May need to allow for jitter
penalty.
# 79
Check SR/ SW vertical eye closure penalty value.
# 81
Check SR/ SW link power penalties.
LR/LW
-----
Obtaining test fibers or equivalent for 1310 dispersion penalty measurement
# 364
power budget, 1310 nm Tables 52- 12, table 52- 14 and tables 52- 15 are not
clear understandable and probably inconsistent.
Send to Serial PMD ad hoc.
# 367,368,130
maximum average transmitter power and receiver overload changed to 0.5 dBm
by #130 and sent to Serial PMD ad hoc for final resolution.
# 89
Check LR/ LW Tx triple trade off table.
# 86
Check LR/ LW Rx sensitivity value. May need to allow for jitter penalty.
# 87
Check LR/ LW stressed Rx sensitivity value: thought very demanding by some.
May need to
allow for jitter penalty.
# 88
Check LR/ LW vertical eye closure penalty value.
# 90
Check LR/ LW link power penalties. Need to allow for reflection noise and,
to be decided,
jitter penalty.
ER/EW
-----
# 363
power budget, 1550 nm Tables 52- 17, table 52- 18 and tables 52- 19 are not
clear understandable and probably inconsistent.
Send to Serial PMD ad hoc.
# 96
Check ER/ EW link power penalties. May need to allow for jitter penalty, to
be decided.
Tuesday's phone conference
---------------------------
Usual coordinates:
15:15 GMT = 4:15 pm BST = 17:15 CET = 11:15 am EDT = 8:15 am PDT,
Tuesday
+1(816)650-0631 Access code 39209
Piers