一种新型的轨到轨高斯函数发生电路

Novel Rail-to-Rail Gaussian Function Generating Circuits

提出了一种轨到轨电压输入、电流输出的高斯函数产生电路的实现方式,它是模糊控制器以及神经网络等电路中的重要部件。通过在传统的高斯函数产生电路的输入级引入PMOS和NMOS差分对,并针对不同输入电压实现独立控制,该电路实现了轨到轨的电压输入,即输入电压达到了电源电压范围。此外,该电路可以根据外加电压调整高斯函数形状,具有匹配误差小等优点。采用无锡上华(CSMC)0.6μm数模混合工艺仿真并流片试验。结果表明,其工作状况良好,在[0,Vdd]的满幅度输入范围内,最大满刻度误差为-2.4%～3.6%。该电路为模糊逻辑(神经网络)系统的硬件实现奠定了良好的基础。

The implementation of a rail-to-rail voltages input,current output Gaussian function generating circuit was proposed.The circuit is the most important component in fuzzy controller and neural network.By introducing the input stage of PMOS and NMOS differential pairs into the traditional Gaussian function generator,and for different input voltages to achieve the independent control,the circuit achieves the rail-to-rail voltage input,that is,the input voltage reaches the power supply voltage range.In addition,the circuit can adjust the Gaussian function according to the shape of the applied voltage,and the circuit has the advantage of small matching error.By Wuxi CSMC 0.6 μm mixed-signal CMOS technology simulation,the experiment results show that the work is good in the full input range of ,the maximum full scale error is between-2.4% and 3.6%.The circuit is a fundamental research for the fuzzy logic（neural network） system hardware implementation.