An Onchip Electrode Impedance Estimation achieving 1.2 dB 3σ-Accuracy with Minimum Hardware Overhead

Neuromodulation has rapidly evolved in recent years and is evaluated for therapeutic treatment of various neurological ailments, including epilepsy and Parkinson’s disease. When transitioning from a measurement laboratory prototype to a biomedical circuit for chronic implantation in humans, safety features and regulations are of utmost importance. One example for a safety-related feature is built-in electrode impedance estimation. Monitoring the electrode condition is crucial for early detection of electrode aging and decay, and allows to adapt stimulation parameters to the changes in the electrode-tissue interface. Prior implementations used time domain analysis or transfer function estimations, which are both bound to a predefined model and vulnerable against noise and disturbers. In this paper, we present an electrode impedance estimation procedure that relies on a simple tone-based estimation to achieve about 1.2 dB accuracy. The estimation is applicable over a wide frequency range with minimum hardware overhead. Compared to our previous implementation, processing complexity is reduced by >100x, and accuracy is improved by 1.6x.