Research Projects

Signal Acquisition Front-end for Implanted Brain-Machine Interfaces

Rikky Muller and Jan M. Rabaey

Gigascale Systems Research Center

The realization of a fully implantable brain-machine interface (BMI) chip will enable life-changing opportunities such as the development of motor prosthetics. The signals of interest, action and local field potentials, run at biological timescales in the range of 1-10kHz and can be as small as 10uV, which presents the challenge of low signal to noise ratio (SNR) at the sensor input. Thus a low-power low-noise signal acquisition front-end is critical to any brain-machine sensor array interface to amplify and convert the neural data for digital signal processing. A significant amount of work has been devoted to this problem over the last decade [1,2]; however, the prior art has relied heavily on analog techniques and passives to perform signal conditioning and filtering, which significantly impacts die area and does not result a scalable solution. We propose an architecture in a fine-line process, which uses feedback from the digital domain to set filter pole locations thereby eliminating the need for the integration of large passive components. Combining over-sampling acquisition and digital signal processing in a reconfigurable system can result in significant implant power and area reduction.

Figure 1: Fig. 1: System Block Diagram

[1]

R. Harrison, C. Charles, “A Low-Power Low-Noise CMOS Amplifier for Neural Recording Applications” IEEE JSSC, Vol. 38: 958-965, Jun. 2003.

[2]

T. Denison, K. Consoer, W. Santa, A. Avestruz, J. Cooley, A. Kelly, “A 2uW 100nV/rtHz Chopper-Stabilized Instrumentation Amplifier for Chronic Measurements of Neural Field Potentials” IEEE JSSC, Vol. 42, No. 12, Dec. 2007.