Design Guidelines for 100 Gbps

news2024/11/14 22:59:58

文章目录

  • Stratix V GT Transceiver Channels
  • CFP2 Host Connector Assembly and Pinout
  • Stratix V GT to CFP2 Interface Layout Design
  • Board Stack Up Dimensions
  • Example Design Channel Performance
  • Simulation Results for Stratix V GT to CFP2 Connector Layout Design

Design Guidelines for 100 Gbps - CFP2 Interface
This document shows an example layout design that implements a 4 x 25/28 Gbps CFP2 module interface that meets the insertion and return loss mask requirements proposed in the working clause draft version 8.0 for CEI-28G-VSR.
The common electrical interface CEI-28G-VSR implementation architecture (IA) for short reach channels is intended for next generation 100 Gbps chip - to - optical module applications. CFP2 is a pluggable optical module that uses CEI-28G-VSR as its electrical interface (as defined by the CFP Multi-Source Agreement (MSA) member companies). CFP2 also defines the mechanical form factor for a 100 Gbps optical transceiver module targeted for Ethernet and OTN (Optical Transport Network) applications.
CFP2 provides an industry standard to develop next generation 100 G interfaces with lower power and greater port density compared to previous generation CFP optical modules.
Note: For more information, refer to the CEI-28G-VSR working clause specification. Document number OIF2010.404.08.
Figure 1: Stratix V GT Device to a CFP2 Pluggable Module Interface on a PCB
在这里插入图片描述

The channel layout on the PCB is optimized in order to meet the strict insertion and return loss masks defined by CEI-28G-VSR.
Refer to the following documents for more information on optimizing your board designs for high speed serial links.
Related Information
• AN529: Via Optimization Techniques for High-Speed Channel Designs
• AN530: Optimizing Impedance Discontinuity Caused by Surface Mount Pads for High-Speed Channel Designs

Stratix V GT Transceiver Channels

Stratix® V GT FPGAs offer four transceiver channels (ATT_TXR[3:0]_P/N and ATT_RXR[3:0]_P/N) that can operate up to 28 Gbps for interfacing with CFP2 or other optical modules.
Figure 2: Top View of 28 Gbps Transmitter and Receiver Channel Locations in Stratix V GT FPGAs
在这里插入图片描述

CFP2 Host Connector Assembly and Pinout

The CFP2 specification defines the mechanical connector requirements for the 104-pin CFP2 connector. The host connector assembly is composed of a female host connector, and a metal connector cover and cage for retention and electromagnetic shielding of the inserted CFP2 optical module.
Figure 3: CFP2 Host Connector Assembly for a 4x25G/28G Module Interface as Defined by the CFP2 Mechanical Specification
在这里插入图片描述

Note: This figure is courtesy of Yamaichi Electronics.
Figure 4: CFP2 Host Connector Pinout for 4x25G/28G Module Interface as Defined by the CFP2 Mechanical Specification
在这里插入图片描述

Figure 5: CFP2 Host Connector Layout Footprint
The high-speed transceiver pins are identified in the following figure to show their position within the connector. Blue pins are the TX transceiver channels and red pins are the RX transceiver channels.
在这里插入图片描述

Stratix V GT to CFP2 Interface Layout Design

The TX and RX channels are connected directly to the CFP2 connector with approximately 5.5 inches of differential trace routing on the top and bottom layer of the board. DC blocking capacitors are included in the optical module for both the TX and RX traces. Nominal trace impedance is controlled at approximately 100Ω differential and the board material used is Panasonic Megtron-6.
Figure 6: Stratix V GT to CFP2 Interface Layout Design Example
The figure shows an example layout design where the green traces are the TX channels routed on the top layer while the orange traces are the RX channels routed on the bottom layer.
在这里插入图片描述

In this example, vias are used for the RX channel breakout at the BGA, and for both the TX and RX channels at the CFP2 connector. To avoid the top layer keep out requirement of the CFP2 metal connector cover assembly, the TX channel routing is switched briefly to the bottom layer and then back to the top layer at the CFP2 connector as illustrated by the circled area in the above figure. Top to bottom routing is used to avoid via stubs.
The BGA pads, signal vias, and CFP2 trace to pad interfaces are large discontinuity sources in the channel. Ansys HFSS (High Frequency Structural Simulator) 3-D field solver simulation is used to optimize the BGA breakout and CFP2 interface design. The trace impedance is kept within ±10% of the nominal 100Ω

Board Stack Up Dimensions

The detailed trace design and board stack up dimensions are shown in the figure below.
Figure 7: Differential Trace Construction and Stack Up Details
在这里插入图片描述

BGA Breakout Optimization
BGA breakout optimization targets both the BGA pads and dog bone vias. A cutout is provided in the reference plane under the BGA pad and large oval via anti-pads are used for better BGA pad and via impedance matching.
Figure 8: BGA Via Breakout Layout Optimization
在这里插入图片描述

Figure 9: TDR of BGA Via Breakout
HFSS simulation results show that the TDR deviation of the BGA escape is maintained within ±10% of the nominal 100Ω channel target impedance.
在这里插入图片描述

CFP2 Interface Optimization
The CFP2 host connector layout optimization reduces the impact of discontinuity at the differential pair to the CFP2 connector interface. A reference plane cutout is provided beneath the connector pads and larger oval anti-pads are used for the signal vias. Four nearby ground return vias are provided to help reduce the connector interface discontinuity.
Figure 10: CFP2 Connector Interface Layout Optimization
在这里插入图片描述

Figure 11: HFSS Simulated TDR of the CFP2 Connector interface
The following figure shows the HFSS simulated TDR results. With the layout optimizations, the TDR deviation due to the discontinuity caused by the via and connector pad is kept within ±10% of the nominal 100Ω target impedance.
在这里插入图片描述

Example Design Channel Performance

The CEI-28G-VSR working clause defines several mask requirements for the channel, including insertion loss, return loss, and differential-to-common mode conversion. Because it is difficult to verify electrical parameters of a full channel in a system, the working clause also defines a Host Compliance Board (HCB) with test points for verifying the host-to-module channel performance at various test points.
Figure 12: Host Compliance Board Measurement Points
在这里插入图片描述

For example, TP1a and TP4a define the measurement points and the associated mask requirements for the host-to-module electrical signal performance for insertion loss, return loss, and differential-to-common mode conversion. For more information about these definitions, refer to the CEI-28G-VSR working clause specification (document number OIF2010.404.08).

Simulation Results for Stratix V GT to CFP2 Connector Layout Design

Ansys HFSS (High Frequency Structural Simulator) simulation results for the insertion loss (SDD21), return loss (SDD11) and differential-to-common mode conversion (SDC11) of the channel with the CFP2 connector included are shown in the following figures. The simulation models the HCB for validating the channel layout against the CEI-28G-VSR defined masks.
Figure 13: Insertion Loss versus CEI-28G-VSR Mask Requirements
The SDD21 resides within the HCB minimum and maximum insertion loss masks as defined by the CEI-28G-VSR specification. This insertion loss meets the complete VSR channel (host board + connector + optical module) mask requirement with ample margin to accommodate the additional loss of an inserted CFP2 optical module. Note that the complete channel with the optical module is not simulated.
在这里插入图片描述

Similarly, the figures below show that the return loss and differential-to-common mode conversion both meet their respective masks as defined by the CEI-28G-VSR requirement.
Figure 14: Return Loss versus CEI-28G-VSR Mask Requirements
在这里插入图片描述

Figure 15: Mode Conversion versus CEI-28G-VSR Mask Requirements
在这里插入图片描述

本文来自互联网用户投稿,该文观点仅代表作者本人,不代表本站立场。本站仅提供信息存储空间服务,不拥有所有权,不承担相关法律责任。如若转载,请注明出处:http://www.coloradmin.cn/o/1243478.html

如若内容造成侵权/违法违规/事实不符,请联系多彩编程网进行投诉反馈,一经查实,立即删除!

相关文章

使用 NVProf 检测 CUDA kernel 的 bank conflict

使用 NVProf 检测 CUDA kernel 的 bank conflict NVProf 指令 使用 NVProf 可以对 bank conflict 进行检测: nvprof --events shared_ld_bank_conflict,shared_st_bank_conflict <app> [args...]其中: --events 选项指定的 shared_ld_bank_conflict,shared_st_bank_c…

【Django使用】10大章31模块md文档,第5篇:Django模板和数据库使用

当你考虑开发现代化、高效且可扩展的网站和Web应用时&#xff0c;Django是一个强大的选择。Django是一个流行的开源Python Web框架&#xff0c;它提供了一个坚实的基础&#xff0c;帮助开发者快速构建功能丰富且高度定制的Web应用 全套Django笔记直接地址&#xff1a; 请移步这…

camera-caps:Jetson设备上的一种实用的V4L2可视化界面

camera-caps&#xff1a;Jetson设备上的一种实用的V4L2可视化界面 github地址是&#xff1a; https://github.com/jetsonhacks/camera-caps 注意&#xff1a;Jetpack5.x需要选择tag 5.x版本

【基础知识】AB软件RSLinx的版本说明

哈喽&#xff0c;大家好&#xff0c;我是雷工&#xff01; 之前对AB的软件了解比较少&#xff0c;在工作中未接触过&#xff0c;最近一次现场勘察时&#xff0c;有很多中控系统都是AB的&#xff0c;借此机会对AB软件有了些许了解。 一、RSLinx是什么软件&#xff1f; RSLinx是…

【LeetCode刷题-回溯】-- 46.全排列

46.全排列 方法&#xff1a;回溯法 一种通过探索所有可能的候选解来找出所有的解的算法&#xff0c;如果候选解被确认不是一个解&#xff0c;回溯法会通过在上一步进行一些变化抛弃该解&#xff0c;即回溯并且再次尝试 使用一个标记数组表示已经填过的数 class Solution {pu…

vue3自定义拖拽指令

<template><div v-move class"box"></div> </template><script setup lang"ts"> import { Directive } from vue const vMove:Directive (el:HTMLElement) >{const mousedown (e:MouseEvent) >{// 鼠标按下const s…

『 C++类与对象 』多态之单继承与多继承的虚函数表

文章目录 &#x1fae7; 前言&#x1fae7; 查看虚表&#x1fae7; 单继承下的虚函数表&#x1fae7; 多继承下的虚函数表 &#x1fae7; 前言 多态是一种基于继承关系的语法,既然涉及到继承,而继承的方式有多种: 单继承多继承棱形继承棱形虚拟继承 不同的继承方式其虚表的形…

redis运维(十八)pipeline

一 pipeline 流水线 说明&#xff1a; 这里讲解的不是jenkins的pipeline流水线这里pipeline: 管道 redis为什么要提供pipeline功能 事务和pipeline ① pipeline的理念 强调&#xff1a;单纯的pipeline跟事务没有关系redis-cli --pipe --> 使用了pipeline机制说明&a…

【Skynet 入门实战练习】游戏模块划分 | 基础功能模块 | timer 定时器模块 | logger 日志服务模块

文章目录 游戏模块基础功能模块定时器模块日志模块通用模块 游戏模块 游戏从逻辑方面可以分为下面几个模块&#xff1a; 注册和登录网络协议数据库玩法逻辑其他通用模块 除了逻辑划分&#xff0c;还有几个重要的工具类模块&#xff1a; Excel 配置导表工具GM 指令测试机器人…

CAD图纸设计在线协同、CAD图纸设计在线协同方案?

CAD图纸设计在线协同、CAD图纸设计在线协同方案&#xff1f; CAD图纸设计在线协同&#xff0c;在企业产品研发效能的提升中发挥着重要作用&#xff0c;技术应用的深入发展为不同场景的协作带来了全新的应用模式&#xff0c;工业设计领域亦是如此。 在CAD图纸设计与管理过程中&a…

RabbitMQ 搭建和工作模式

MQ基本概念 1. MQ概述 MQ全称 Message Queue&#xff08;[kjuː]&#xff09;&#xff08;消息队列&#xff09;&#xff0c;是在消息的传输过程中保存消息的容器。多用于分布式系统之间进行通信。 &#xff08;队列是一种容器&#xff0c;用于存放数据的都是容器&#xff0…

【分布式】小白看Ring算法 - 03

相关系列 【分布式】NCCL部署与测试 - 01 【分布式】入门级NCCL多机并行实践 - 02 【分布式】小白看Ring算法 - 03 【分布式】大模型分布式训练入门与实践 - 04 概述 NCCL&#xff08;NVIDIA Collective Communications Library&#xff09;是由NVIDIA开发的一种用于多GPU间…

SQL进阶学习

1.[NISACTF 2022]join-us sql报错注入和联合注入 过滤&#xff1a; as IF rand() LEFT by updatesubstring handler union floor benchmark COLUMN UPDATE & sys.schema_auto_increment_columns && 11 database case AND right CAST FLOOR left updatexml DATABA…

CLion安装与配置教程

目录 一、下载并安装CLion1、下载1、官网&#xff1a;2、注意&#xff1a; 2、安装1、下载完成后&#xff0c;直接点击安装包安装&#xff0c;即可。2、开始安装&#xff0c;然后下一步3、可以在此处自定义地址&#xff0c;然后下一步4、根据系统版本选择&#xff0c;然后下一步…

Linux:虚拟机安装Ubuntu系统

一、下载Ubuntu 地址&#xff1a;https://cn.ubuntu.com/download/desktop 二、安装 以上配置完成后&#xff0c;点击完成按钮&#xff0c;接下来就是一段较长时间的等待安装过程。 安装完成后&#xff0c;还有一些系统性配置。 系统配置非常简单&#xff0c;全部next即可。…

开源 GPU池化软件 | (AI人工智能训练平台、AI人工智能推理平台)

GPU池化软件 | (AI人工智能训练平台、AI人工智能推理平台) 讨论群v:&#x1f680;18601938676 一、AI人工智能开发-------------面临的问题和挑战 1. GPU管理难题 1.1 资源管理难&#xff1a;算力资源昂贵&#xff0c;但是缺乏有效管理&#xff0c;闲置情况严重。 1.2 用户…

【uniapp】uniapp开发小程序定制uni-collapse(折叠面板)

需求 最近在做小程序&#xff0c;有一个类似折叠面板的ui控件&#xff0c;效果大概是这样 代码 因为项目使用的是uniapp&#xff0c;所以打算去找uniapp的扩展组件&#xff0c;果然给我找到了这个叫uni-collapse的组件&#xff08;链接&#xff1a;uni-collapse&#xff09…

Django 入门学习总结4

视图是Django应用程序在Python语言中提供特定的方法并对应于有特定的模板的网页。网页的页面通过视图的方式进行跳转。 在投票系统中&#xff0c;有四个视图&#xff1a; 首页视图&#xff0c;显示最新的问题列表。细节视图&#xff0c;显示问题文本&#xff0c;通过表单可以…

【标注数据】labelme的安装与使用

这里写目录标题 下载标数据 下载 标数据 打开自动保存 创建矩形

FreeRTOS的并行与并发思考

FreeRTOS的任务触发是由滴答时钟触发SysTick中断来触发调度器执行或阻塞或挂起和切换任务的。 首先是任务的并发能力&#xff0c;FreeRTOS的任务执行是基于全抢占调度机制&#xff0c;任务优先级按在就绪列表中由高到低排布&#xff0c;系统首先执行最高优先级任务&#xff0c;…