24 Channel dual-band wireless neural recorder with activity-dependent power consumption

摘要

A complete light weight 24 channel wireless neural recording system enabled by a custom IC is presented. By transmitting only the action potential (AP) data, the system power consumption is scaled based on the activity of the neural signal. By detecting an AP event prior to digitization, the ASIC can reduce the power consumed in all subsequent stages and yet conserves the entire AP shape without affecting the classification performance. The complete 2.5 cm3 head stage weighing below 7 g can operate for 3 h. The ASIC achieves <3.5 µVrms noise (both in AP and LFP band) and >70 dB CMRR.