Scholarlink

<p>Neuromorphic engineering has garnered significant interest for its potential in creating energy-efficient and highly parallel computing systems. One of the key components of such systems is the neuron circuit, especially Axon Hillock, which plays a vital role in signal integration and propagation. This paper explores the design of Axon Hillock (A-H) neuron circuits using a 28&#xa0;nm Fully Depleted Silicon on Insulator (FD-SOI) MOSFET due to its advantages over Bulk CMOS. The original A-H neuron circuit undergoes two distinct modifications. In the first modification, original differential amplifier’s functionality is replaced with the inherent inverter threshold voltage, resulting in a reduced transistor count, lower power consumption of 26.3&#xa0;µW, and an improved frequency of 9.18&#xa0;kHz. The second modification involves replacing the differential amplifier with a low-threshold 2-transistor (2-T) based differential circuit, achieving a nearly 50% reduction in power consumption (16.8&#xa0;µW) and a 1&#xa0;kHz frequency boost (9.78&#xa0;kHz) compared to the original A-H neuron circuit. Further, the third modified circuit eliminates the differential amplifier, membrane capacitance (C<sub>mem</sub>), and other control transistors, transforming it into a low-power variant. This circuit has a power consumption of 6.9 pW and an increased frequency of 63.34&#xa0;kHz, nearly 18-fold increase compared to the original A-H neuron circuit.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Implementation and analysis of axon hillock neuron circuits using 28 nm FD-SOI MOSFET: original design and modifications

  • 期刊论文

  • 发表日期：

    2025年7月19日

查看全文
  • Sasi Kiran Suddarsi,
  • Sandeep Moparthi,
  • Gopikant Kumar,
  • Harika Ganta,
  • Saranya Sri Peddapudi,
  • Harshitha Goru,
  • Gopi Krishna Saramekala

摘要

Neuromorphic engineering has garnered significant interest for its potential in creating energy-efficient and highly parallel computing systems. One of the key components of such systems is the neuron circuit, especially Axon Hillock, which plays a vital role in signal integration and propagation. This paper explores the design of Axon Hillock (A-H) neuron circuits using a 28 nm Fully Depleted Silicon on Insulator (FD-SOI) MOSFET due to its advantages over Bulk CMOS. The original A-H neuron circuit undergoes two distinct modifications. In the first modification, original differential amplifier’s functionality is replaced with the inherent inverter threshold voltage, resulting in a reduced transistor count, lower power consumption of 26.3 µW, and an improved frequency of 9.18 kHz. The second modification involves replacing the differential amplifier with a low-threshold 2-transistor (2-T) based differential circuit, achieving a nearly 50% reduction in power consumption (16.8 µW) and a 1 kHz frequency boost (9.78 kHz) compared to the original A-H neuron circuit. Further, the third modified circuit eliminates the differential amplifier, membrane capacitance (Cmem), and other control transistors, transforming it into a low-power variant. This circuit has a power consumption of 6.9 pW and an increased frequency of 63.34 kHz, nearly 18-fold increase compared to the original A-H neuron circuit.