Capacitive tuning of thyristor oscillators enables neuron-like signal amplification

Self-oscillator devices based on a negative differential resistance (NDR) offer a promising foundation for oscillator-based computing—an emerging computational paradigm that exploits the synchronization, phase dynamics, and collective behavior of coupled nonlinear oscillators for efficient pattern recognition, temporal coding, and signal processing—as well as for implementing spiking neural networks. In this work, we present a compact, thyristor-based two-terminal device that exhibits an ultrasmooth, hysteresis-free NDR response, enabling precise and robust control over complex temporal dynamics. Through a combination of experimental measurements and numerical simulations, we demonstrate that the system supports a tunable Hopf bifurcation, with oscillatory behavior modulated by input current and capacitance. As capacitance increases, the oscillations transition smoothly from sinusoidal to relaxation-type waveforms. The underlying bifurcation structure is analyzed analytically and visualized through dynamical trajectories. Importantly, when operated near the bifurcation threshold, the device exhibits stochastic resonance, amplifying weak periodic signals in the presence of noise, and yielding a 6.8-dB improvement in signal-to-noise ratio. These results establish a neuromorphic-compatible, low-power architecture that bridges nonlinear device physics with functional applications in temporal computation and brain-inspired hardware.