Signal integrity has become a major problem in digital IC design. One cause of this problem is device scaling which results in a sharp reduction of supply voltage, creating stringent noise margin requirements to ensure functionality. Reductions in feature size also result in increased clock speeds leading to many different high frequency noise producing components. As on-chip area increases to allow for more computational capability, so does the amount of digital logic to be placed, magnifying the effects of noisy interconnect structures.
Supply noise, modeled as AV = Ldi/dt , is caused by rapid current spikes during a rise or fall time. Decoupling capacitors often fill empty on-chip space for the purpose of limiting this noise. This work introduces a novel methodology that attempts to quantify and locate decoupling capacitors within a power distribution network. The bondwire attached on the periphery of the face of the die is taken to be the dominant source of inductance. It is shown that distributing capacitance closer to the switching elements is most effective at reducing supply noise.
A chip has been designed using TSMC 90 nm technology that implements the ideas presented in this work. Simulation results show that noise fluctuations are high enough such that random placement of decoupling capacitance is not effective for large digital structures. The amount of interconnect generated on-chip noise increases with area, resulting in the need for an optimal decoupling scheme. As scaling continues, supply voltages and noise margins will decrease, creating the need for a robust decoupling capacitance methodology.
Library of Congress Subject Headings
Digital integrated circuits--Power supply; Capacitors--Design and construction; Systems on a chip; Electronic noise
Computer Engineering (MS)
Department, Program, or Center
Electrical Engineering (KGCOE)
Bohannon, Eric Scott, "On-Chip Digital Decoupling Capacitance Methodology" (2007). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus