当前位置:首页 > 报告详情

DC25_SLIDES_Track12_TutorialChannelDeEmbeddingMethodology_Choi_V2.pdf

上传人: S** 编号:1240833 2026-05-16 49页 5.84MB

1、Information Classification:GeneralWelcome to ConferenceJanuary 2830,2025Santa Clara Convention Center1ExpoJanuary 2930,2025Information Classification:GeneralChannel Embedding/De-Embedding Methodology for Rx Stress Testing Using Fast-Edge Tx Waveform of BERTYoonman Choi,SK hynixKwang-rok Kim,Keysight

2、2Information Classification:GeneralImageImage SPEAKERSYoonman ChoiSenior App Engineer,SK hynixY|Senior application engineer of MPHY/PCIe/SATA for storage device solutions.Located in Seongnam-Si,Gyenggi-do,Korea.Kwangrok KimApplication engineer,KeysightKwang-|Kwang-Rok Kim is an Advanced Solution eng

3、ineer at Keysight Technologies.He received his Bachelors degree in electronic engineering.He joined Keysight Digital team in 2019 and works on High speed digital application and Solution Engineering in Digital.Prior to 2019 he worked in the areas of system integration,programming development,automat

4、ion.3Information Classification:GeneralIntroduction4Information Classification:GeneralProblem Statement5 Most commercially available standardized high-speed serial IOs define the receiver compliance or conformance tests using the worst case channel scenario which have the maximum channel loss.To eva

5、luate the overall receiver performance with the worst case channel,the Bit Error Rate Tester(BERT)is a commonly used for receiver compliance or conformance tests as a test signal source.But,even when the DUT can tolerate the DC/AC input voltage tolerance(ITOL)or receiver jitter tolerance(JTOL)with t

6、he worst case channel,if we replace the test signal source from BERT to real silicon,we find a margin issue or complete fail results.If we replace the signal source from BERT Tx to the actual host/device silicon Tx,is it possible to guarantee the channel Rx margin with the same test channel?Or if we

word格式文档无特别注明外均可编辑修改,预览文件经过压缩,下载原文更清晰!
三个皮匠报告文库所有资源均是客户上传分享,仅供网友学习交流,未经上传用户书面授权,请勿作商用。
1. **问题背景**:BERT测试与实际硅片Tx存在接收机余量不匹配问题,需解决测试环境与真实信道的差异。 2. **核心方法**:通过嵌入/去嵌入TTC(时间转换器)调整BERT Tx边沿速率,匹配硅片Tx波形(如-2dB@11.648GHz)。 3. **关键数据**: - MIPI MPHY HS-G5B(23.296Gbps)测试中,TTC级联增加插入损耗(0/-2/-4/-6dB),影响眼图高度(如957.5mUI→894.2mUI)。 - 硅片Tx边沿慢于BERT,需额外-2dB损耗补偿以匹配眼图。 4. **测试建议**: - 使用目标信道模型(如MIPI CH1,-10.55dB@11.64GHz)校准接收机抖动容限(JTOL)。 - 结合可变ISI设备动态调整损耗,验证CDR与Rx均衡器性能。 5. **结论**:需考虑TTC损耗和Tx边沿差异,确保最坏情况信道余量。
**BERT与硅片差异?** **如何校准测试通道?** **接收器容限如何验证?**
客服
商务合作
小程序
服务号
折叠