神经网络并行算法型CMOS模数转换器设计

基于神经网络和开关电容技术提出了一种行算法型ＣＭＯＳ模数转换器电路结构，文中详细分析了电路的工作原理，并采用通用电路模拟程序进行了仿真，结果表明转换功能正确，输入电压递增和递减均无错码，转换过程中不需要复位，该转换器充分发挥开关电容精确处理和神经网络并行处理的优点，因而精度高，速度快，而且所需元件少，工艺完全与常规ＣＭＯＳ兼容，是一种有巨大发展潜力的新型ＣＭＯＳ转换器。

DESIGN OF A PARALLEL ALGORITHM ADC BASED ON NEURAL NETWORKS

A parallel algorithm CMOS ADC is described which is based on an neural network and switched capacitor technology. The operation principle is discussed in detail. The converter has been simulated by using PSPICE circuit simulator. The results show that the converter codes correctly for both the increase and decrease of the input voltage and needs no reset during conversion. By taking advantage of the accurate processing of switched capacitor circuits and the parallel processing of neural networks, the converter is suitable for high speed and high resolution application. Meanwhile, it requires fewer components, and can be implemented in conventional CMOS technology. It is a new mode CMOS ADC with a great potential in the future.