{"title":"基于113gb /s PAM-4收发器的1.1 pj /b/Lane 1.8 tb /s芯片及均衡策略降低分数间隔0.5 ui ISI","authors":"G. Gangasani;A. Mostafa;A. Singh;D. Storaska;D. Prabakaran;K. Mohammad;M. Baecher;M. Shannon;M. Sorna;M. Wielgos;P. Jenkins;P. Ramakrishna;U. Shukla","doi":"10.1109/LSSC.2025.3526877","DOIUrl":null,"url":null,"abstract":"This letter uses 113-Gb/s PAM4 transceiver in 5-nm CMOS to demonstrate a 1.8-Tb/s chiplet, over die-to-die extremely short-reach (XSR) intrapackage links, in an 8-port configuration. The 16-channels range from 1 to 12 dB of loss at <inline-formula> <tex-math>$F_{\\textrm {baud}}/2$ </tex-math></inline-formula>. The chiplet performance over these channels is better than <inline-formula> <tex-math>$\\textrm {BER}\\lt 10^{-9}$ </tex-math></inline-formula>, while consuming <1.1-pJ/b power and 0.22-mm2 area per lane. The performance targets are achieved using an transceiver equalization strategy which minimizes 0.5-UI ISI by design in the data path and using a LUT-based TX FFE-3 for signal equalization and envelope adaptation.","PeriodicalId":13032,"journal":{"name":"IEEE Solid-State Circuits Letters","volume":"8 ","pages":"33-36"},"PeriodicalIF":2.2000,"publicationDate":"2025-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"A 1.1-pJ/b/Lane, 1.8-Tb/s Chiplet Using 113-Gb/s PAM-4 Transceiver With Equalization Strategy to Reduce Fractionally Spaced 0.5-UI ISI in 5-nm CMOS\",\"authors\":\"G. Gangasani;A. Mostafa;A. Singh;D. Storaska;D. Prabakaran;K. Mohammad;M. Baecher;M. Shannon;M. Sorna;M. Wielgos;P. Jenkins;P. Ramakrishna;U. Shukla\",\"doi\":\"10.1109/LSSC.2025.3526877\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"This letter uses 113-Gb/s PAM4 transceiver in 5-nm CMOS to demonstrate a 1.8-Tb/s chiplet, over die-to-die extremely short-reach (XSR) intrapackage links, in an 8-port configuration. The 16-channels range from 1 to 12 dB of loss at <inline-formula> <tex-math>$F_{\\\\textrm {baud}}/2$ </tex-math></inline-formula>. The chiplet performance over these channels is better than <inline-formula> <tex-math>$\\\\textrm {BER}\\\\lt 10^{-9}$ </tex-math></inline-formula>, while consuming <1.1-pJ/b power and 0.22-mm2 area per lane. The performance targets are achieved using an transceiver equalization strategy which minimizes 0.5-UI ISI by design in the data path and using a LUT-based TX FFE-3 for signal equalization and envelope adaptation.\",\"PeriodicalId\":13032,\"journal\":{\"name\":\"IEEE Solid-State Circuits Letters\",\"volume\":\"8 \",\"pages\":\"33-36\"},\"PeriodicalIF\":2.2000,\"publicationDate\":\"2025-01-06\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"IEEE Solid-State Circuits Letters\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://ieeexplore.ieee.org/document/10829856/\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q3\",\"JCRName\":\"COMPUTER SCIENCE, HARDWARE & ARCHITECTURE\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"IEEE Solid-State Circuits Letters","FirstCategoryId":"1085","ListUrlMain":"https://ieeexplore.ieee.org/document/10829856/","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"COMPUTER SCIENCE, HARDWARE & ARCHITECTURE","Score":null,"Total":0}
A 1.1-pJ/b/Lane, 1.8-Tb/s Chiplet Using 113-Gb/s PAM-4 Transceiver With Equalization Strategy to Reduce Fractionally Spaced 0.5-UI ISI in 5-nm CMOS
This letter uses 113-Gb/s PAM4 transceiver in 5-nm CMOS to demonstrate a 1.8-Tb/s chiplet, over die-to-die extremely short-reach (XSR) intrapackage links, in an 8-port configuration. The 16-channels range from 1 to 12 dB of loss at $F_{\textrm {baud}}/2$ . The chiplet performance over these channels is better than $\textrm {BER}\lt 10^{-9}$ , while consuming <1.1-pJ/b power and 0.22-mm2 area per lane. The performance targets are achieved using an transceiver equalization strategy which minimizes 0.5-UI ISI by design in the data path and using a LUT-based TX FFE-3 for signal equalization and envelope adaptation.