چكيده به لاتين
Rapid development of microelectronics during the past years allowed the emergence of high-performance implantable biomedical microsystems (IBMs). Neural recording microsystems are the most important parts of implantable biomedical devices. A neural recording microsystem is consists of neural sensory circuit, multiplexer, analog to digital converter and wireless power and data telemetry stages. In this thesis the concentration is on designing and simulating neural sensory circuit and multiplexer blocks to be used in neural recording microsystems.
Weak neural signals must be amplified before can be processed further. The low noise and low power amplifier is the main building blocks of the neural sensory circuits. In this thesis, the closed loop capacitive feedback structure is used to design neural amplifier, in order to provide sufficient gain, excellent linearity and high common mode and power supply rejection ratios. One of the most common OTAs for designing neural amplifiers is current reuse OTA, due to its low power consumption, input referred noise and area. However, it suffers from a few restrictions such as low current efficiency and reduction of the transconductance by increasing the input differential voltage. Thus in this thesis, these restrictions have been overcome by applying some modifications while maintaining the proper features of this structure.
The proposed amplifier is designed and simulated in 180nm CMOS standard technology with a 1.8V supply voltage. The proposed OTA in open loop configuration has a gain of 63dB, input referred noise of and total power consumption of 270nW. To achieve the desired frequency range of the spike signals and also to cancelling DC offset of the electrodes, this OTA is designed and simulated in the closed loop capacitive feedback structure. In this case, the closed loop amplifier yielded a midband gain of about 40dB and 3-dB bandwidth from lower than 1Hz to 7kHz; also applying the input differential voltage of results in THD less than 0.12% at this frequency range. Furthermore, NEF and PEF of the proposed neural amplifier are respectively 1.01 and 1.83.
The multiplexer stage is the other important parts of sensory circuits. Multiplexers are formed by analog switches. The accuracy of these switches is limited by various factors such as charge injection, clock feedthrough, off-state leakage current and nonlinear distortion. In this thesis, a small and simple switch with minimized charge injection and clock feedthrough errors is presented. Moreover, off-state leakage current of this switch is reduced significantly by an innovative method.
Simulation results show that the proposed switch achieves signal to noise and distortion ratio (SNDR) of 81.04dB, effective number of bits (ENOB) of 13.16, total harmonic distortion (THD) of -89.87dB and spurious-free dynamic range (SFDR) of 98.55dB for a 1.220703125kHz sinusoidal input of 1.8V peak-to-peak amplitude at 62.5kHz sampling rate with a 1.8 supply voltage. Also, its leakage is negligible for input variations from 0.2V to 1.2V; at the worst condition it is less than 60fA. As a result, the proposed switch can significantly improve the dynamic and static performances of a sampled-data system.
