Power Conditioning and Stimulation for Wireless Neural Interface ICs

Power Conditioning and Stimulation for Wireless Neural Interface ICs by William James Biederman III Doctor of Philosophy in Engineering Electrical Engineering and Computer Sciences University of California, Berkeley Professor Jan M. Rabaey, Chair Brain machine interfaces have the potential to revolutionize our understanding of the brain, restore motor function, and improve the quality of life to patients with neurological conditions. In recent human trials, control of robotic prostheses has been demonstrated using micro-electrode arrays, which provide high spatio-temporal resolution and an electrical feedback path to the brain. However, after implantation, scar tissue degrades the recording signal-to-noise ratio and limits the useful lifetime of the array. This work presents two systems which utilize wireless techniques to mitigate this e↵ect and create high-density, long-term interfaces with the human brain. A wirelessly powered 0.125mm 65nm CMOS IC integrates four 1.5μWamplifiers (6.5μVrms input-referred noise with 10kHz bandwidth) with power conditioning and communication circuitry. Multiple nodes free-float in the brain and communicate via backscatter to a wireless interrogator using a frequency-domain multiple access communication scheme. The full system, verified with wirelessly powered in vivo recordings, consumes 10.5μW and operates at 1mm range in air with 50mW transmit power. A 65nm CMOS 4.78mm neuromodulation SoC integrates closed loop BMI functionality on a single IC which can be arrayed on a wireless sub-cranial platform. The IC consumes 348μA from an unregulated 1.2V supply while operating 64 acquisition channels with epoch compression (at an average firing rate of 50Hz) and engaging two stimulators (with a pulse width of 250μs/phase, di↵erential current of 150μA, and a pulse frequency of 100Hz). Compared to the state of the art neural SoCs, this represents the lowest area and power for the highest integration complexity achieved to date.