A multi-channel neural recording system with a highly area- and energy-efficient analog front end (AFE) has been developed at the University of Michigan. Neural recording systems that provide simultaneous monitoring over a large number of channels within a small area, while maintaining high signal quality, are desirable for comprehensive neuroscience research. Extant recording systems feature on the order of a few hundred parallel recording channels; this number is expected to reach over a thousand within a decade. However, reducing the power and area in designs with over 1000 parallel recording channels is critical, and is an open problem.
The neural recording architecture developed at the University of Michigan reduces power and area, and achieves an energy-area product of 4.84fJ/C-s·mm2, the smallest reported to date. The energy- and area-efficient modular analog front-end (AFE) architecture incorporates Δ-modulated ΔΣ (Δ-ΔΣ) signal acquisition for 1024-channel brain activity monitoring platforms. The AFE employs spectrum-equalizing and continuous-time ΔΣ quantization to make use of the inherent spectral characteristics of brain signals. Further, the dynamic range of the neural signals is compressed by 27dB (spectrum equalization). Overall, the design demonstrates low area and high energy-efficiency, reduced crosstalk between channels, and a scalable architecture.
- Neural recording systems
- Low area and improved energy efficiency
- Relaxed crosstalk between channels
- Support for easy expansion