[History and development of optical Ethernet] "ER8" aiming for a reach of 40 km at 800 G, adopting MZM and paralleling two DSPs for 400 G [Net new technology]-INTERNET Watch
Using MZM can cover a distance of 40km
What is the actual silicon result for this? The figure below is the data of the paper "Silicon Photonics Applications for 5G and Data Centers" presented at OFC 2021.
![[History and Development of Optical Ethernet]](https://website-google-hk.oss-cn-hongkong.aliyuncs.com/drawing/article_results_9/2022/3/9/5be82f52c1dd194eebf616c7bc11c0d8_1.jpeg "[History and development of optical Ethernet] Aiming for reach 40km at 800G Adopting")
As the title suggests, the content is about manufacturing a 400G transceiver using Silicon Photonics and examining its characteristics.
The lead author of this paper is Mr. Rang Chen Yu of SiFotonics, who is the presenter of the presentation. Sample #2 is a conventional transceiver that conforms to the "100G ER1-40" spec specified by the 100G Lambda MSA.The EYE Pattern in the graph shows the waveform after implementing "400G-DR4" using that Silicon Optics and performing PAM4 modulation, and it shows that the Eye Height is properly taken. being appealed.
There is no detailed description in the paper, but in the explanation of the graph above, "We have also tested transmission dispersion tolerance with our internal developed silicon photonics MZ modulators PIC, as shown in Fig. 2, which show good margin vs. 100G ER1-40 specification by 100G Lambda MSA.", and compared to 100G ER1-40 of 100G Lambda MSA, it is said that there is a large margin (low dispersion penalty) when using MZM, and MZM As long as you use , the graph is probably intended to cover a transmission distance of 40km.
However, as mentioned earlier, when using the formula for chromatic dispersion defined by the ITU, -60 to -37 ps/nm is actually 1308 to 1310 nm, so here If 8 wavelengths are passed through , the interval is about 0.29 nm.
This figure of -60 to -37 ps/nm is shown in the graph above.This is a wavelength density that exceeds DWDM every 0.8 nm, and as expected, the feasibility is low. Too much. Therefore, by widening the limit of chromatic dispersion a little more and making it a form of eight wavelengths between 1302 and 1310nm, it is possible to implement it realistically.
In this case, the chromatic dispersion will be -85 to 37 ps/nm, and the total will be about 122 ps/nm, so it seems necessary to work hard on the DSP side to cover this.