SPICE memristor model based on estimate from measured data

Purpose The real production memristor devices produce a modeling challenge due to having a memristor variance and henceforth a measured difference between breadboard made circuits and LTSPICE or MATLAB simulated circuits with classical models, resulting in variances owed to difference in existing models and real hardware. When a new memristor is studied, usually no model is provided and it’s useful if there is a tool that automatically update models with production parameters. The purpose of this paper is to implement a procedure in MATLAB that takes measured data, approximates parameters and provides a MATLAB and LTSPICE models for precise representation of real memristor devices. Design/methodology/approach The optimal values of production level memristor model’s coefficients can be estimated for various existing models by applying the measured voltage-current relationship and by using optimization procedure to match the coefficients to existing model. With graphical user interface (GUI) in MATLAB environment a user can select measured data and which model to be used as some are good for high frequency and others for low or mid frequency memristors. Findings The analyses, which were performed in MATLAB and LTSPICE, validate the efficiency and accuracy of the proposed memristor model matching procedure. The analysis case utilizes a comparison with some very commonly used standard and modified memristor models. Originality/value A GUI with optimization procedure for parameter estimation of real-world production memristors into simple memristor models or with parasitic parameters is applied in MATLAB and then transferred in LTSPICE with amplitude-frequency responses and voltage-current relations analyzed for different frequencies.