电流模式MOS　A／D转换神经网络的研究

A CURRENT-MODE MOS A/D CONVERTER BAS ON NEURAL NETWORK

从Ａ／Ｄ转换基本规则出发，提出了一种电流模式ＭＯＳＡ／Ｄ转换神经网络，并对该网络进行了分析研究，导出了确定权值的公式，网络权值易存贮且调节简单。用ＰＳＰＩＣＥ对４－ｂｉｔＡ／Ｄ转换网络进行了仿真，并对采用分立元件构成的３－ｂｉｔＡ／Ｄ转换网络进行了实验，仿真与实验结果均表明该网络可正确进行Ａ／Ｄ转换。

Using A/D conversion rule a current-mode MOS A/D converter based on neural net-work is presented in this paper.The weights can be stored and adjusted easily.The operationof the network and determination of various weights are shown,The operation of a 4-bit A/Dconverter is simulated with PSPICE.A3-bit A/D conversion circuit is implemented by usingdiscrete MOSFETs.The resuIts obtained,both for computer simulations and for experimentalmeasurements,show that this network gives correct outputs for both increasing and decreas-ing analog inputs and no reset is needed during conversion.