The main goal of this work was to design an amplifier which is able to amplify signals of neural sensors, taking the challenging properties of these sensors and the measurable signals of the brain themselves (including impedance, signal levels, bandwidth etc.) into accout. To design an amplifier that has noise and offset performances sufficiently small, it is needed to use some type of dynamic offset/noise cancellation. In my thesis I have chosen the method of chopper stabilization to meet these specifications.
First research has been done on the topic of neural/brain signals, and the sensors that are used to monitor them, with emphasis on the silicon deep-brain sensors fabricated by MTA MFA. Electric model (eg. equivalent circuit) of the latter was used in the simulations of the amplifier.
I have reviewed the literature of the chopper technique, its principle, and the errors that appear in the practical use of the method. I have also made a short overview of the simulation problems of such circuits, and the way to simulate them using the Cadence SpectreRF circuit simulator.
Due to the specifications a high gain amplifier is needed. To make the frequency compensation of the amplifier easier, a simple version of the so-called Multipath feedforward amplifier topology was implemented. The paper describes the design of the two stage operational amplifier (chopped folded cascode fully differential first stage, symmetrical OTA for signal adding with Class-AB output stage in the second stage) in detail, and the theoretical background that was used during the design. The performance of the circuit was investigated by simulating the gain, noise, offset, CMRR, PSRR and input common mode range. I have also verified that the amplifier works properly in the specificated temperature and supply voltage range in all technology corners (PVT simulations).