상세 보기
Data-Driven Jitter Estimation in PLL-to-HBM3E Chip-Level Clock Path
- Park, Taeho;
- Kim, SoYoung
SCOPUS
0초록
This work presents a data-driven, on-chip approach for period jitter prediction in a PLL-to-HBM3E clock distribution path. Conventional EDA-based jitter analysis relies on multi-stage workflows and long runtimes, limiting rapid design iteration. Extensive transient SPICE simulations were performed to generate a comprehensive dataset. A linear regression model was then trained to estimate on-die jitter directly from power-supply noise features such as dV/dt, amplitude, and process corner. The proposed model achieved an R2 value of 0.69 and an RMSE of 0.53 ps, while reducing the simulation turnaround time from several days to nearinstantaneous prediction. These results demonstrate that machine learning can effectively complement EDA-based transient analysis, providing a faster yet physically consistent method for jitter estimation. © 2026 IEEE.
키워드
- 제목
- Data-Driven Jitter Estimation in PLL-to-HBM3E Chip-Level Clock Path
- 저자
- Park, Taeho; Kim, SoYoung
- 발행일
- 2026
- 유형
- Conference paper
- 저널명
- Final Program - 2026 Asia-Pacific International Symposium and Exhibition on Electromagnetic Compatibility, APEMC 2026
- 언어
- ENG
- 출판사
- Institute of Electrical and Electronics Engineers Inc.