Low-Power Circuits and Systems Design for Data Acquisition and Transmission in a Wireless Cortical Implant

Minimally invasive monitoring of the electrical activity of specific cortical areas using implantable microsystems offers the promise of diagnosing neurological diseases, as well as detecting and identifying neural activity patterns which are specific to a behavioral phenomenon. Multi-channel recording in-vivo impose critical constraints on circuits and systems design in order to comply with severe safety requirements. The system power consumption should be sufficiently small in order to enable autonomous and battery-less operation as well as to limit the temperature increase due to operation of the implant. Moreover, small area and volume of the implant are critical constraints to minimize issues associated with surgery operation and implantation. In this research, low-power circuits and systems techniques for data acquisition and transmission in wireless multi-channel cortical implants are presented. Wireless power transmission is carried out using an inductive coupling technique along with fully on-chip low-drop-out (LDO) voltage regulation. Stable operation over a wide range of load conditions, and fast load and line regulation are the main design issues of LDO regulators, which are addressed by proposing a novel compensation methodology, and a power supply rejection ratio (PSRR) boosting technique. Low-noise operation of the analog front-end is obtained by introducing new design techniques at circuit and system levels. The partial OTA sharing technique is proposed as a circuit-level approach which results in a significant reduction of power dissipation as well as silicon area, in addition to a very low noise efficiency factor (NEF). The effect of mismatch on crosstalk between channels, trade-off between noise and crosstalk, and nonlinearity of the amplifiers are theoretically analyzed and confirmed by measurement results. Three different system architectures are presented, which preserve the temporal information of the recording sites by avoiding channel multiplexing. A 16-channel neural recording system is presented, which uses an oversampling delta modulator as a dedicated ADC per channel. The oversampling delta modulator not only improves the system level NEF, but also provides in-site compression of the slow varying neural signal. The fabricated prototype consumes 220 μW from a 1.2 V power supply and achieves an input-referred noise equal to 2.8 μVrms. The application of algebraic Walsh-Hadamard coding in multi-channel recording systems is investigated by developing a 16-channel prototype. The linear and orthogonal combination of channels provided by coding, maps the spacial information of the channels to the temporal information of a superposed signal, and enables parallel recording from multiple channels using a single ADC. Moreover, this technique improves the spacial resolution of the recording sites by moving the shared signal processing hardware to the outside of the sensor plane. A fabricated chip supports a sensor pitch equal to 250 μm, consumes 359 μW from a 1.2 V power supply, and achieves an input-referred noise equal 4.1 μVrms. Finally, a system on a package (SoP) is presented which consists of a 64-channel neural recording system named Neuro+II, and an impulse radio ultra wideband (IR-UWB) transmitter. Neuro+II hosts the power conversion and voltage regulation blocks, the analog/mixed-mode front-end unit, and the digital baseband processing module. A dynamic power scaling technique is presented which enables 20.4% reduction in power consumtion of the analog/mixed-mode front-end. Neuro+II consumes 3.26 mW and achieves a power dissipation density equal to 13 mW/cm2. An IR-UWB transmitter is presented as an up-link communication module of the Neuro+II. An eight pulse-position modulation (8-PPM) scheme is implemented using a novel all-digital delayed-locked-loop (DLL) circuit, which offers better spectral compliance with USA Federal Communication Commision (FCC) regulations. A symmetric pulse combining technique is proposed to reduce the number of power amplifier elements by half, which enhances the tuning range capability of the transmitter. The fabricated transmiter consumes 540 μW and achieves an energy efficiency of 45 pJ/bit with an output power measured at -26 dBm. Continuous improvements in the field not only support research in the life science do- main, but also enables the clinical treatment of some diseases and extends the application field of such systems from clinical experiments to in-house treatments and ambulatory monitoring.