The purpose of this work is to compare dynamical modes in the ensemble of hardware electronic generators of neuron-like activity with dynamical modes in SPICE simulator and mathematical model in order to detect whether the difference in oscillation amplitude, form and bifurcation value of control parameter between the hardware generators and simulation is a result of model imperfection, or this difference can be explained by features of the used electronic elements.

Models and methods. Mathematical models, SPICE simulations and three hardware copies of the tunable generator are considered. The dependence of the excitation threshold and the oscillation amplitude on the control parameter is determined for different nonlinearities due to the number of diodes in the feedback loop. The mutual information function is used to compare the waveform.

Results. It is shown that the existing differences can be fully explained by standard variations in the parameters of semiconductor components and other circuit elements used for the construction of electronic neurons. In this case, the simulation model can be considered as one of the generators, the parameters of which could be precisely controlled, and its components had zero tolerances.

Conclusion. Modern simulation models are able to give a fairly good description of a full-scale experiment, it is impossible to distinguish the time series of the simulator from the experimental ones; at the same time, the experimental implementations themselves may differ due to random variations in the properties of the components.