A Sub-500 mu W Interface Electronics for Bionic Ears

Ulusan H., Muhtaroglu A., KÜLAH H.

IEEE ACCESS, vol.7, pp.132140-132152, 2019 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 7
  • Publication Date: 2019
  • Doi Number: 10.1109/access.2019.2940744
  • Journal Name: IEEE ACCESS
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.132140-132152
  • Keywords: Fully implantable cochlear implant, bionic ear, neural stimulation, ultra-low power, current-mode, COCHLEAR IMPLANT, NEURAL STIMULATION, WAVE-FORMS, MICROPHONE, SENSOR, RANGE, CHIP
  • Middle East Technical University Affiliated: Yes


This paper presents an ultra-low power current-mode circuit for a bionic ear interface. Piezoelectric (PZT) sensors at the system input transduce sound vibrations into multi-channel electrical signals, which are then processed by the proposed circuit to stimulate the auditory nerves consistently with the input amplitude level. The sensor outputs are first amplified and range-compressed through ultra-low power logarithmic amplifiers (LAs) into AC current waveforms, which are then rectified through custom current-mode circuits. The envelopes of the rectified signals are extracted, and are selectively sampled as reference for the stimulation current generator, armed with a 7-bit user-programmed DAC to enable patient fitting (calibration). Adjusted biphasic stimulation current is delivered to the nerves according to continuous inter-leaved sampling (CIS) stimulation strategy through a switch matrix. Each current pulse is optimized to have an exponentially decaying shape, which leads to reduced supply voltage, and hence similar to 20% lower stimulator power dissipation. The circuit has been designed and fabricated in 180nm high-voltage CMOS technology with up to 60 dB measured input dynamic range, and up to 1 mA average stimulation current. The 8-channel interface has been validated to be fully functional with 472 mu W power dissipation, which is the lowest value in the literature to date, when stimulated by a mimicked speech signal.