Kim K H, Kim S J
School of Electrical Engineering, Seoul National University, Korea.
Med Biol Eng Comput. 2000 Jul;38(4):469-72. doi: 10.1007/BF02345018.
A noise performance design method for the pre-amplifiers of an active neural probe is given. The on-chip circuitry of the active neural probe consists of CMOS devices that show high-/low-frequency noise, so that the device noise can become dominant. Analysis of the signal-to-device-noise ratio (SDNR) for the CMOS source follower buffer and two-stage differential voltage amplifier is given. Closed-form expressions for the output noise power are derived and exploited to tailor the parameters that are controllable during circuit design. The output SDNR is calculated considering the real extracellular action potentials, the electrode-electrolyte interface and the noise spectrum of CMOS devices from typical foundries. It is shown that the output device noise power can be much higher than the output signal power if the devices at the input stage of the pre-amplifier are made as small as given fabrication technology permits. Quantitative information of the circuit parameters to achieve an SDNR higher than 5 for neural spikes with 60 microV amplitude are provided for both pre-amplifier types.
给出了一种有源神经探针前置放大器的噪声性能设计方法。有源神经探针的片上电路由呈现高频/低频噪声的CMOS器件组成,因此器件噪声可能占主导地位。给出了CMOS源极跟随器缓冲器和两级差分电压放大器的信噪器件比(SDNR)分析。推导了输出噪声功率的闭式表达式,并利用这些表达式来调整电路设计中可控制的参数。考虑实际的细胞外动作电位、电极 - 电解质界面以及典型晶圆厂CMOS器件的噪声谱来计算输出SDNR。结果表明,如果前置放大器输入级的器件尺寸缩小到给定制造技术允许的最小值,输出器件噪声功率可能会远高于输出信号功率。针对两种类型的前置放大器,都提供了对于幅度为60微伏的神经尖峰实现高于5的SDNR所需的电路参数的定量信息。