[802.3ae_Serial] Clause 52 vertical eye closure penalties
Clause 52'rs -
I had a couple of discussions with an esteemed veteran of the group on
how the vertical eye closure penalty is calculated. He was very helpful,
but I am still somewhat confused. I can find no documentation on how it
is determined, and it does not show up directly in the spreadsheet.
Stressed Rx sensitivity
* When converted to OMA, stressed sensitivity = P_launch_min -
P_channel. Revision 3.0 included the Rx eye offset penalty, but I recall
that was removed in St. Louis, and this change should appear in revision
3.1. No issue.
Vertical eye closure
* P_isi is included. No issue.
* P_mpn, P_rin, and P_mn are not included. These are "noise"
penalties (not deterministic like P_isi) and therefore not subject to
the eye closure definition in Figure 52-15. However, since they are not
included in the calculation, the assumption must be that these penalties
are included as part of the test source. I am feeling a little mixed
about counting on that...
* P_cross is dominated by baseline wander. It does not appear to
be included in the calculation. This assumes it is an Rx effect, which I
feel a little mixed about. A Tx could have baseline wander due to AC
coupling and/or an APC loop that is too fast.
* Unallocated margin - should this be in the stressed test values,
or does it show up in the Tx triple tradeoff curves?
* Finally, calibration of the stressed eye is to be done with a
SONET filter at 7500 MHz, yet P_isi is calculated in the spreadsheet
using 8250 MHz. Shouldn't the vertical eye closure value be larger for
the 7500 MHz filter, so that is provides the value in the spreadsheet
under the 8250 MHz scenario?
Receive sensitivity in OMA
* Nothing to specifically say here, except that if P_mpn, P_rin,
P_mn, P_cross, and Margin are not included appropriately in the stressed
Rx test, then maybe we still need this spec.
My main motivation here is more to understand how the vertical eye
closure penalty is calculated as part of my review, not to challenge
anything (not yet, anyway). I would like to hear responses to these
questions and concerns.
Thanks in advance,
Tom Lindsay
Stratos Lightwave
425/672-8005