Hi Mark et all,
since I’ll not be able to make the meeting next week, but I see my test results cited into
yu_3cu_01_0919
slide 3 and 4, is important to clarify some points that should have driven misleading conclusions by Yu Xu, Jialong Shuai and Xinyuan Wang.
In slide 3:
- The transmitter is an uncooled SiP, NOT a cooled one. This is a QSFP28 module designed to be compliant to
100GBASE-FR1 applications.
- Indeed the test experiment reported in
mazzini_3cu_adhoc_082119 was done by heating and cooling the same tranmitter, to sweep across 8nm range (slide 2) and cover the min-max chromatic dispersions foreseen for 400GBASE-LR4.
- The T effect are already included then, is NOT a SiP TX at room temperature, but it has been heat and cooled even
beyond the T range for the module (slide 2).
In slide 3, a negative margin for TX OMA-TDECQ of -0.36dB is reported in slide 3, by speculating
on my comment about power launch (slide 4) to say that the TX OMA-TDECQ is not adequate to ensure 400GBASE-LR4 spec.
- Since the focus of the experiment was to show TDECQ and TDECQ-10*Log(Ceq) compliance as per comments in my last
slide. Then NOT to show its full compliance to all other TX parameters.
- Since the TX is designed for 100GBASE-FR1 application, it has been tuned to the right power that allow such transmitter
to sit in the mid of the foreseen OMA range for THAT PMD, which is between required to be between -0.2 to 4.2 OMA.
- Having low TDECQ after 2kms propagation, also 100GBASE-FR1 OMA-TDECQ requirements are met by this TX then.
- A simple tuning of laser current for the 400GBASE-LR4 will adjust to the right OMA and OMA-TDECQ region then if
this PMD will be .
In slide 3, a margin of 0.4dB* is supposed to be not enough because ‘An additional penalty
is expected to be introduced at high temperature’.
- When the transmitter in the experiment is ALREADY at high T (even beyond expected T range as said).
You should firstly fix these errors when referring to my presentation and results.
Regards
Marco
*Then one can still say that there’s ‘only 0.41dB margin’, when this would occur when:
- the worst case L0 fiber dispersion is present in the link (hard to find in the field, supposed to be =< 2% as per
stassar_cu_adhoc_041719 )
- the
exact maximum link length (it will be just percentage of the 400GBASE-LR4 deployed links, 1 link over 10 ? it’s probably less than that )
- the TX at the edge of the CWDM grid at that particular WL corner, which depends on corner T too (looking at some
internal distribution on DVT data on similar PMDs this quite a small percentage << 1/1000, stay safe with 1/1000)
With these assumption makes 0.2*0.1*0.001 =
2 links over 1 million
- to drop the link we of course need to have
all other TX and RX parameters at the worst case (very hard to quantify < .... 1/1000 ?)
With these assumption makes 0.2*0.1*0.001*0.001 =
2 links over 1 billion over which in real life we’ll have to rely on ‘just’ this 0.41dB of margin... still respecting any opinion, I would design for that.
From: Mark Nowell (mnowell) <00000b59be7040a9-dmarc-request@xxxxxxxx>
Sent: viernes, 6 de septiembre de 2019 13:15
To: STDS-802-3-100G-OPTX@xxxxxxxxxxxxxxxxx
Subject: [802.3_100G-OPTX] P802.3cu web page and meeting time update
Dear Colleagues,
The contributions for next week’s meeting have been posted here:
http://www.ieee802.org/3/cu/public/Sept19/
The meeting will start @ 9am rather than 8am on Monday morning.
We are in Grand 5 meeting room.
Safe travels.
Mark
To unsubscribe from the STDS-802-3-100G-OPTX list, click the following link:
https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-100G-OPTX&A=1
To unsubscribe from the STDS-802-3-100G-OPTX list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-100G-OPTX&A=1