New 32-channel event-based AFE ASIC slashes power for brain-computer interfaces
Adaptive delta encoding boosts data compression 10x, enabling always-on neural monitoring under 1 µW per channel.
Low-power event-based analog front-ends are critical for building efficient neuromorphic signal processing systems. A new chip from researchers at the University of Zurich and ETH Zurich (Giacomo Indiveri’s group) delivers a 32-channel AFE ASIC optimized for biomedical signal acquisition and encoding. The chip uses dual-mode encoding: Pulse Frequency Modulation (PFM) and an adaptive Asynchronous Delta Modulator (aADM). The aADM encoder provides an auto-scaling mechanism that adapts the encoding data-rate based on the input signal envelope in real-time, enabling very high data compression for low-power information transmission.
Fabricated in a 180 nm CMOS process, the ASIC offers a highly configurable interface compatible with state-of-the-art Spiking Neural Network (SNN) neuromorphic processors. This approach directly addresses the power bottleneck in wireless neural recording for brain-computer interfaces. By compressing neural signals on-chip before transmission, the system can sustain long-term, always-on monitoring without draining batteries. The 32 independent input channels allow simultaneous recording from multiple sites, making it suitable for high-density neural probes and wearable bio-sensors.
- Fabricated in 180 nm CMOS, the ASIC has 32 independently programmable input channels with dual-mode encoding.
- The adaptive Asynchronous Delta Modulator (aADM) auto-scales data rate in real-time, achieving high compression for low-power wireless transmission.
- Designed to interface directly with Spiking Neural Network (SNN) processors, enabling efficient end-to-end neuromorphic processing for brain-computer interfaces.
Why It Matters
This ultra-low-power AFE chip could make continuous, wireless neural monitoring practical for real-world brain-computer interfaces and wearable health devices.