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

[802.3ae_Serial] From serial PMD call 5 Mar part 1: options for TTO simplification




Present
-------
Piers Dawe		Agilent SPG
Raj Savara		Network Elements
Stretch Camnitz	Agilent T&M
Tom Lindsay		Stratos
Petar Pepeljugoski  IBM
Greg LeCheminant	Agilent T&M
Adam Healey		Agere
Peter Ohlen		Optillion
Mike Dudek		Cielo
David Kabal		Picolight

Triple Trade off complexities: we now have TTO for Tx OMA and for TDP,
except 10GBASE-E.  Is this really necessary?

Group felt that individual products would state wavelength, spectral width,
OMA power and TDP as necessary; the datasheet values would be compliant to
just a few cells of the triple trade off.  Similarly for 10GBASE-E,
datasheet would declare OMA power and TDP, among other specs.  So users of a
particular manufacturer's model will not be exposed to the full range of
choice in the standard.

The discussion centred on 10GBASE-L.  The other case, 10GBASE-S, has
slightly different issues but a similar range of options.

Options for simplification as my email
http://www.ieee802.org/3/10G_study/public/serial_adhoc/email/msg00577.html .
Further options discovered:

8.	Delete the cells at the long wavelength extreme.
pro: no-one uses them.  con: they show the way to e.g. 40GE by link
aggregation and CWDM.

9.	Increase the quantization step from 0.1 to 0.2 dB
pro: reduces number of different cells by ~1/2.	  con: wastes 0.1 dB.

10.	Delete one or more of the wide spectral width columns
pro: some people know they don't need e.g. 0.15 to 0.2 nm spectral width.
con: others may need it.

11.	Specify (OMA -TDP +correction for predicted attenuation).
pro:	Flexible, similar to 10GBASE-E, doesn't waste up to 0.4 dB, avoids
inaccurate (for 10GBASE-L) dispersion penalty calculation in the model.
con:	On the call we feared this would be like option 4 with another
dimension, i.e. as complex as what we are trying to simplify.  However, we
can simply write the equation,
OMA -TDP -0.0036*(lambda - 1310) > spec value, very much like 10GBASE-E.  No
table needed!

We timed out on the discussion.  I'll try to complete the minutes of the
meeting (other subjects) tomorrow.

There will not be a phone meeting next week; we have a real meeting.
Commenting deadline: "ballot for IEEE P802.3ae/D4.1 Recirculation officially
opens on Monday, 25-February-2002, and it closes at 11:59 p.m.(Eastern Time)
on Saturday, 09-March-2002."

Piers