A Neural Recorder with Mixed Signal Feedback Loop for a Spectral Separation of Spikes and LFP

Abstract:
A front-end circuit for neural recording applications with a mixed signal feedback loop for a spectral separation of neural action potentials (spikes) and Local Field Potantial (LFP) is presented in this paper. Spectral separation of neural signals has the advantage, that the data rate, the TX power and the analog-digital-converter (ADC) resolution can be reduced. But, all analog filter structures for spectral separation consume a lot of power and are therefore not very efficient. A mixed signal feedback structure has the advantage that the filter works in the digital domain. Thus, the implementation is very efficient and can be used for neural recording applications. The presented structure consists of a fully differential low-noise amplifier (LNA) which has a very good noise efficiency factor. A feedback structure splits the neural signal into the sub signals. Therefore, a digital integrator and a σ δ converter for the digital to analog conversion (DAC) are realized in the feedback path. The whole front-end circuit consumes 21.9muW and has a input referred noise of 3.8muVrms with band pass corners of 450Hz and 6.5kHz. It is simulated in 0.18micrometer high voltage CMOS technology