| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
Dear 802.3 Colleagues, At the Working Group meeting later today, 802.3cn Task Force will be submitting 400GBASE-ER8 standard for approval. Based on the data presented in 802.3cn TF meetings, as well as related data presented in 802.3cu TF meetings, this PMD is
will be very difficult to build because there is little if any manufacturing margin to key specs. This undermines the entire premise for the PMD which is low cost, therefore it fails to meet the BMP Criteria (Broad Market Potential). Given that there is an
alternative in Coherent, the PMD will meet the NMP Criteria (No Market Potential). No evidence was submitted to the TF that there is even single End User that actually wants a very expensive 40km PMD. In contrast, in 802.3cu, End Users were very clear that
arbitrary reaches like 10km don’t matter, and cost is everything. Because of the proliferation of SMF TFs, it has become impossible to stay on top all the different technical activities. With the proposal to split 802.3ct, we will now have two IMDD SMF projects and two Coherent SMF projects. Ideally there
should only be one 802.3 project per media. At most two should be allowed, in this case one for IMDD LR and ER PMDs and one for Coherent ZR PMDs. Doubling this to four, leads to specifications blazing through without the appropriate review. Further by having competing solutions to the same problem in 802.3cn, there is an inherent conflict of interest. Many of those in favor of Coherent, have been actively speaking out against IMDD for longer reaches at conferences and other
forums. This means they have little motivation to invest their time to carefully review the IMDD spec. which will compete directly with what they are interested in. All the SMF IMDD LR and ER specs. should have been in a single 802.3cu project, and all Coherent
ZR should be in a single 802.3cn project. Because we don’t have that, the shifting mix and match of projects means important specs. are not reviewed.
The 400GBASE-ER8 spec. has only been out this year. There is no need to rush it through. My recommendation to the 802.3 Working Group is to ask 802.3cn to take another meeting cycle to carefully review the 400GBASE-ER8 spec and come back
to the Working Group with demonstrated BMP.
Chris From: Chris Cole Dear Colleagues, The issue being raised with the 400GBASE-ER8 proposed specification is not narrowly whether there have been any dispersion measurements, but rather that the specification has not received the kind of scrutiny that is typical of 802.3 standards.
An example of appropriate level of scrutiny is what happened this year in 802.3cu, with the 400GBASE-LR4 proposal on CWDM4 grid supporting 10km reach. This proposal was vigorously challenged, which led to multiple iterations of measurements,
broad analysis of those measurement by many individuals, and a review of the underlying assumptions behind the reach objective so that the specification actually met the Broad Market Potential criteria. This was in the best tradition of 802.3, and resulted
in a revised adopted specification with 6km reach objective. Most importantly, it addressed real End User requirements, which are first and foremost low cost, which in turn will lead to a highly successful PMD.
This is in contrast to what has happened in 802.3cn with respect to 400GBASE-ER8 proposal on LWDM grid supporting 40km reach. In below emails, favorable measurements are listed which show that the 40km reach can just be met, just as there
are favorable measurements in 802.3cu showing that 10km reach can just be met. What is missing is a range of measurements, including using different technologies. Just as LR4 on CWDM4 grid for 10km is a stretch, so ER8 on LWDM for 40km is a stretch. Which
misses the whole point of creating an IMDD spec. for >10km. There is a solution for reaches beyond IMDD capability, and that is Coherent. At some reach, in order for IMDD to be viable, it has to be substantially lower cost than Coherent, otherwise there is
no motivation for End Users to proliferate optics types. They have no choice but to use Coherent. They have a choice whether to use ER8. The 400GBASE-ER8 proposed specification is not low cost because there is little manufacturing margin. The only presented measurements have been favorable ones using EML TX with
positive {changed from “negative” per comment in response email by John Johnson} chirp, which compensates for SMF dispersion at short wavelengths. This means that other technologies, in particular MZ TX may not work. To deliberately exclude Silicon
Photonics from an IEEE specification is bizarre to say the least, given the huge industry investment in Silicon Photonics technology. This investment likely exceeds investment in any other technology. 802.3 specs are not written to favor just one technology,
because it is recognized that picking technology winners should not be part of the standards process.
The End Users have been very clear. Meeting arbitrary past reach numbers which made technical sense at lower data rates is much less important than lowest possible cost for the rate of interest. At 400G, 40km ER makes no more sense than
10km LR. http://www.ieee802.org/3/cu/public/Sept19/cole_3cu_01b_0919.pdf#page=3 A low cost ER8 spec. needs to be targeted at a lower reach, for example 25km, so that there is ample manufacturing margin for TX power and penalty specs. and multiple technologies can be used to implement the TX. The low cost required of
an IMDD optic to be competitive with a Coherent solution will not be achieved if there is little manufacturing margin.
The right next step is to go back and subject the current specification to vigorous review, and to obtain accurate End User feedback on what’s more important to them in a >10km IMDD PMD, 40km or low cost. Otherwise the ER8 40km spec. will
not only not meet the Broad Market Potential Objective, but will quickly have no market at all as it’s supplanted by Coherent. This guidance maybe have to be provided to 802.3cn TF by the 802.3 Working Group. Thank you Chris From: Hideki Ooe <ooe-hideki@xxxxxxxxx>
Dear John, Below presentation addresses TDECQ down to -200 ps/nm dispersion (with T-spaced 5-tap FFE), shown in page 5. As usual to EML, it shows negative CD penalty down to -150 ps/nm. http://www.ieee802.org/3/B10K/public/18_09/jackson_b10k_01_0918.pdf Regards, ------------------------------------------------- Hideki Ooe <ooe-hideki@xxxxxxxxx> Transmission Devices Laboratories Sumitomo Electric Industries From: Brian Welch (bpwelch) <00000e3f3facf699-dmarc-request@xxxxxxxx>
John, Thanks for sending, I hadn’t seen these. However in looking at them I don’t see that any TDECQ measurements were taken. Sone_b10k_01a_1117 does indicate some CD penalties, but it does so using a T/2
spaced 17-tap FFE as the receiver, which is quite a bit different that the reference receiver in use in a TDECQ measurement. Do you know of any TDECQ measurements taken (with the prescribed reference equalizer)? Also, given that there are negative CD penalties down to as low as -150 ps/nm suggests that the transmitters in use have quite a bit of chirp. Do you know if measurements were performed for any un-chirped
transmitters? Thanks, Brian From: John DAmbrosia <jdambrosia@xxxxxxxxx>
All, Yesterday an issue was raised regarding dispersion, and there was an inquiry regarding any presentations on the topic had been heard. My thanks to Peter Stassar, who did the digging through the Study
Group archives to find the following presentations that had explored assessing the dispersion penalty at high negative dispersion of around -200ps/nm. As I recall now, this was a key issue for the Study Group to resolve before it finally adopted the 40km
objectives and could answer the Technical Feasibility Criteria. SG material. November 2017: http://www.ieee802.org/3/B10K/public/17_11/sone_b10k_01a_1117.pdf January 2018: http://www.ieee802.org/3/B10K/public/18_01/yamamoto_b10k_01a_0118.pdf March 2018 http://www.ieee802.org/3/B10K/public/18_03/yu_b10k_01c_0318.pdf Regards, John D’Ambrosia Chair, IEEE P802.3cn Task Force To unsubscribe from the STDS-802-3-B10K list, click the following link:
https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-B10K&A=1
To unsubscribe from the STDS-802-3-B10K list, click the following link:
https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-B10K&A=1
To unsubscribe from the STDS-802-3-B10K list, click the following link:
https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-B10K&A=1 To unsubscribe from the STDS-802-3-DIALOG list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-DIALOG&A=1 |