| Hi Takayama-san,
The short answer is YES. But in 802.3 there are also cases where different PMDs share the same PCS/PMA. For example, -SR, LR, -ER, CR, -KR, etc PMDs share same PCS/PMA, but also includes other layers in between of FEC, etc. to meet the specific requirements of each application.
In case of .3cz (also applicable to .3dh), as I explained last week in the interim meeting, there are functional requirements like OAM and EEE that are of particular interest in the automotive application. These requirements forced .3cz to make modifications with respect to the BASE-R PCS/PMA. Specific time slots were allocated in the transmit frame structure in order to transport end to end the information belonging to the PHD where OAM a and other information for PHY control, dependability and capacities negotiation is contained. For a EEE that allows big power saving in RX side, other changes of how the LPI is encoded and wake is signaled were needed to introduce. In addition, data-aided timing recovery and equalization are supported because specific features in several state diagrams and scrambler, which were also explained last week. The last point is key to support better sensitivity required by higher insertion loss automotive grade connectors.
In BASE-AU, PCS, PMA and PMD are designed as a whole to meet the application specific requirements. As I also explained last week, functional requirements of .3cz and .3dh are the same. The link budget requirements are also similar. So, IMO, the application specific requirements are the same in both projects.
Best regards,
Rubén Pérez-Aranda
Hi, Ruben, Thank you for the information. I get the idea that "connecting BASE-SR PMDs to BASE-U PCS/PMA and viceversa is something that does not make sense and result, at least for me, is undefined and out of the scope of the activities of .3dh project." from the previous paragraph. PCS, PMD and PMD are designed as a whole to meet the application specific requirement, aren't they? Best Regards, Takayama First of all let me clarify about PMD. PMD is not just wavelength, but a sublayer with specific service interface, TX and RX characteristics, nominal rate, definition of optical parameters and measurements methods, fiber optic cable characteristics, etc. The simplest point to check is the service interface. This needs to be compatible with the PMA to which is going to be connected. If you read subclause 112.2 of 25GBASE-SR PMD, you will read that the PMD in the receive direction continuously sends a bit stream to the PMA corresponding to the signals received from the MDI. However, if you read 166.6.1.2, you will read the PMD in the receive direction conveys to the PMA the relative amplitude of the optical signal received by the PMD at the MDI. As you can see there is a subtle but important difference. In clause 112, the information embedded within the optical signal received from MDI are decided by the PMD RX. However, in case of clause 166, PMD does not decide / decode, only translate relative amplitude, so PMA RX can implement clock and data recovery that may implement equalization. Other very important points are the TX and RX characteristics, and definition of parameters and test methods. The TX/RX characteristics of the BASE-SR PMD assume a PMD that operates consistently with service interface. The TX/RX characteristics of BASE-AU PMD assume PMD will be connected to PMA with equalization (see reference receiver defined for TDFOM and SRS calibration that includes MMSE-DFE). Therefore, connecting BASE-SR PMDs to BASE-U PCS/PMA and viceversa is something that does not make sense and result, at least for me, is undefined and out of the scope of the activities of .3dh project. So, let me reformulate the question in different way: What will happen if BASE-U PCS&PMA is combined with a BASE-U compatible PMD that uses 850 nm wavelength? Is it feasible to produce a PMD that operates at 850nm and is compatible with a BASE-U PCS/PMA? My opinion as PHY vendor is that it is technically feasible to produce compatible PMDs that operates at 850nm and meet all the TX and RX characteristics of BASE-AU (except wavelength center) for 25 Gb/s operation. It is demonstrated in the slides 4, 5 and 6 of https://www.ieee802.org/3/dh/public/Sep_2022/perezaranda_3dh_01_2209_vcsels.pdf, which I will present during next ad-hoc meeting. 850nm VCSEL of Vendor D meets the TX characteristics for 25GBASE-AU between -40C and +125C (of course, it is not the case of all the other tested 850 VCSELs). The link budget would be OK and this implementation would be compatible with my proposal for .3dh. Slides 8, 9 and 10 show that same conclusion does not apply for 50 Gb/s PAM4, which reinforces the decision made of 980nm in .3cz project. However, as I explained many times, TX characteristics are only one side of the coin, the wear-out reliability is the other side. Second part of that presentation is about reliability, and conclusions for me are the same as before: 850nm does not meet reliability for the considered mission profile. This mission profile is consistent with a qualification AEC-Q100 grade 2 (max ambient ECU temperature of 105C), where we assume, based on practical experience, max back-side temperature of 125C. In the connection point where the fiber is connected to PHY, fiber itself has to support around 125C. This is a very easy requirement for glass fiber. But it also can be fulfilled with lenses (e.g. EBO) so fiber is not directly exposed to the highest temperature. Based on figure of slide 30, it is easy to verify that reliability would be improved 1 order of magnitude if we consider 85C max ambient temperature (105C back-side). So, IMO from this figure it is straightforward to conclude that PHYs using 850nm could be qualified AEC-Q100 grade 3. Should we consider different temperature grade, e.g. grade 3, for electronics components to used with GI-POF?. In that case I think 850nm may be reliable enough. In my opinion, changes in components between .3cz and .3dh should only rely on GI-POF in order to avoid confusion in the market. Can be used max temperature of GI-POF as distinct characteristics of the medium that limit the qualification grade of the electronics? What is that max temperature limit?
I would like to know what will happen if BASE-U PCS&PMA is combined with 850 nm PMD from BASE-SR. Since you separated VCSELs presentation from BASE-U PSC&PMA proposal, there may be no big issue other than link budget. Today, my straw-poll to adopt BASE-U PCS and PMA sublayers defined in P802.3cz/D3.01 for P802.3dh resulted in that most of the people think more information is needed to make decisions. I would like to receive feedback from you about which topics you consider more information and details are needed. I will do my best to prepare specific contributions that can help to you in understanding that this proposal is the most suitable one for this project. I appreciate your questions and feedback.
To unsubscribe from the STDS-802-3-25GAUTO-POF list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-25GAUTO-POF&A=1
To unsubscribe from the STDS-802-3-25GAUTO-POF list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-25GAUTO-POF&A=1
|