Logo image
標準CMOS製程之雜訊可調變電晶體直流特性量測與模型建立
Thesis

標準CMOS製程之雜訊可調變電晶體直流特性量測與模型建立

王銘鋒
Masters, 國立清華大學, 電機工程學系
2015

Abstract

仿神經 低頻雜訊 雜訊可調變 Neuromorphic Low-frequency noise Noise adaptable
Neurons encode and transmit information by changing its membrane potentials. The random opening and closing of ion channels in the process of membrane potential changing contributes to the 1/f noise spectrum. The 1/f noise caused by the random opening and closing of ion channels and which caused by the surface trap in the transistors are very similar. The 1/f noise is found to play a beneficial role rather than harmful role for neural communication. As the CMOS technology progressing, the transistor noise increases dramatically, and disturbs the accuracy of traditional VLSI signal processing. Studying how neural systems process the signal in 1/f noise may improve VLSI design. Adding noise adaptable transistors into VLSI neuron model can model the 1/f noise in the neurons. Then studying how the effect caused by these noises contributes to study neuron systems in noisy environments. So how to use the noise adaptable transistor is become very important. This thesis chooses three kinds of transistors using CMOS general process and the noise adaptability has been improved. Without additional process steps, they can be added to the existing neuron model. Then analysis the structure and the DC current characteristics of these transistors and build the DC current model.

Metrics

1 Record Views

Details

Logo image