The closed-loop neural stimulation system's biggest concern is an artifact with more than tens of mV occurs due to stimulation. Previous studies have digitized the input signals through the analog-to-digital converter (ADC) after using amplifiers to measure tens of μV neural signals. Since the amplifier reduces the input range, the amplifier's output is saturated when a large stimulation artifact appears. This chip design uses the 2nd-order continuous delta-sigma modulator (DSM) to measure signals without saturation even if stimulation artifacts are entered with the neural signal. Overall, circuit structures were designed with a focus on stable operation even if a sudden large signal came in. Also, it can quickly track the sudden change of signals by adding an auto-ranging algorithm. We present 16 channel neural recording chip with a 65-nm CMOS process and the entire chip area is 1 mm² with 49 μW power consumption. Input-referred integrated noise from dc to 500 Hz was 0.89 μVrms, and more than 93 dB input dynamic range were guaranteed.