A Fully-Integrated Many-Electrodes Pulsed-Voltage Control Architecture for Arbitrary-Waveform Neural Stimulation With High Energy Efficiency

Deeply implantable neural stimulation calls for architectural solutions that are small, efficient and flexible, and that can stimulate many electrodes. To this end, this paper proposes the use of a pulsed (chopped) voltage stimulator, implemented using a low-complexity switch network with fully adaptive real-time control. The fully-integrated control architecture guarantees current waveform reconstruction and charge balancing by continuously monitoring the charge delivered to the electrode, thus offering robustness towards power-supply voltage and electrode impedance variations. The architecture has a high energy efficiency across the entire output operating range. The output waveform is generated in charge samples (slices) of controlled amount; by controlling these slices properly, the desired arbitrary stimulation waveform is constructed. A voltage monitoring circuit is used to apply active charge balancing; the duration of the balancing phase is adjusted by varying the number of charge samples. The feasibility of the architecture is demonstrated with a chip prototype manufactured in a <roman xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">180 nm</roman>, <roman xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1.8 V/5 V</roman> CMOS process, and has an area of only <roman xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0.027 mm</roman><sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><roman>2</roman></sup> per non-multiplexed stimulator channel. Across the entire output operating range, the experimental validation of the prototype demonstrates a source energy efficiency that is up to 35 % better than previously published implementations. The results show that this architecture is a viable solution for next-generation systems for neuromodulation and closed-loop neural monitoring.