SESSION TA 10
SALON 9, Tues, 8
8:30 AM


CLOCK GENERATION AND DISTRIBUTION

Chair: J. Maneatis, JGM Enterprises, Redwood City, CA
Associate Chair: K. Donnelly, Rambus, Inc., Mountain View, CA


10.1 A 1.3 Cycle Lock-In Time Non-PLL/DLL Jitter
Suppression Clock Multiplier Based on Direct
Clock Cycle Interpolation for "Clock on Demand."
8:30 AM
T. Saeki, M. Mitsuishi, H. Iwaki, M. Tagishi
NEC Corp., Kawasaki, Kanagawa, Japan

A 1.3 cycle lock-in time non-PLL/DLL clock multiplier is based on direct clock cycle interpolation with an array structure of short-circuit-current-suppression interpolators. The circuits used in 622MHz clock and data recovery achieve 40ps peak-to-peak jitter and satisfy the ITU-T G.958 jitter tolerance specification.


10.2 A Digitally-Controlled PLL with Fast Locking Scheme for Clock Synthesis Application
9:00 AM
I-C. Hwang, S-W. Kim
Korea Univ., Sungbuk-gu, Seoul, Korea

A digitally-controlled phase-locked loop (DCPLL) in 3.3V 0.6 um triple-metal CMOS achieves fast acquisition employing digital frequency difference detector (DFDD). The prototype DCPLL provides 136ps peak-to-peak cycle-to-cycle jitter at 400MHz (VCO at 800MHz). The acquisition time is measured as 16 cycles (640ns) using the 25MHz reference.


10.3 An Eight Channel 36GSample/s CMOS Logic Analyzer
9:30 AM
D. Weinlader, R. Ho, C-K. Yang 1, M. Horowitz
Stanford Univ., Stanford, CA
1Univ. of California, Los Angeles, CA

A timing analyzer in 0.25 um CMOS uses an oversampling input channel. On-chip compensation corrects both static phase errors in the sampling clocks and sampler input referred offsets. The analyzer achieves 27.8ps resolution with 0.39ps/mV noise sensitivity. A technique for minimizing timing uncertainty due to cycle-to-cycle clock jitter is presented.

BREAK 10:00 AM

10.4 On-Chip Inductance Modeling of VLSI Interconnects
10:15 AM
X. Qi, B. Kleveland, Z. Yu, S. Wong, R. Dutton, T. Young 1
Stanford Univ., Stanford, CA
1Synopsys Corp., Mountain View, CA

Modeling of on-chip inductance captures 3D geometry and process technology effects. Analytical formulae suitable for design are used to estimate inductance of test structures. S-parameter characterization to 10GHz agrees with simulations and analytical modeling. Consideration of substrate effects results in an 18% reduction of wire inductance for wire spacing >40 um.


10.5 Active GHz Clock Network using Distributed PLLs
10:45 AM
V. Gutnik, A. Chandrakasan
Massachusetts Institute of Technology, Cambridge, MA

A distributed clock network alleviates problems of skew and jitter in GHz-clock-speed microprocessors. A chip with a 4x4 array of PLLs in standard 0.35 um single-poly CMOS achieves lock at 1.3GHz. The VCO, phase detector, and loop filter have small-signal stability despite stringent large-signal (initial locking) constraints. The PLLs each take <4000 um2 and consume 25mW at 3V.


10.6 Clock Generation and Distribution for a
IA-64 Microprocessor
11:15 AM
S. Rusu, S. Tam
Intel Corp., Santa Clara, CA

The clock generation and distribution for the first IA-64 microprocessor achieves low skew over a large die using distributed programmable deskew units. Local skew control compensates for load mismatches and within-die process variations, as well as temperature and voltage gradients, with 12.5ps resolution. Debug features include on-die clock shrink and control over deskew settings.

CONCLUSION 11:45 AM