Monte+Carlo+Sensitivity+Analysis

=Monte Carlo Sensitivity Analysis of a BCI EEG Low Noise Amplifier= toc Written by Chris Nguyen

1. Circuit Operation
In the construction of a BCI system, a low noise amplifier as well as an A/D convertor is required to acquire the biopotential from an EEG electrode. This sensitivity analysis will consider two designs for low noise amplifiers. The first design is a low drain current low noise amplifier designed by C. M. Horwitz and published in the IEEE Transactions on Biomedical Engineering. [1] The second design is the amplifier portion of a low noise, low-power EEG acquisition node for scalable BCI designed by Thomas J. Sullivan, Stephen R. Deiss, Gert Cauwenberghs, and Tzyy-Ping Jung from the Department of Biology in the Institute of Neural Computation at University of California, San Diego.

2. Design Specifications Analyzed
All simulations where run using a Monte Carlo Sensitivity Analysis technique with resistor and capacitor component tolerances of +/- 10%.

A. Low Drain EEG Amplifier [1]
The first design considered was the Low Drain EEG Amplifier shown in Figure 1.

B. A Low Noise, Low-Power EEG Acquisition Node for Scalable Brain-Machine Interfaces [2]
The second design considered was the Low Noise, Low-Power EEG Amplifier shown in Figure 2. This design was simulated because of the poor results yielded in the first design.

3. LTSpice Simulation of Nominal Performance Design
The nominal performance design for both circuits was tested under the desired voltage range for external BCI measurements, with the input voltage ranging from -80mV --> 40mV.

A. Low Drain EEG Amplifier
This design was simulated in LTSpice, but the simulation yielded an inverse affect of amplification by reducing the output current by a magnitude of 9 orders.

**﻿****B. A Low Noise, Low-Power EEG Acquisition Node for Scalable Brain-Machine Interfaces**

This design was simulated in LTSpice, but the simulation yielded a redundant affect of amplification, simply by acting as a buffer and passing the signal through.

**4. Monte Carlo Sensitivity Analysis**
The Monte Carlo Sensitivity Analysis was run for both designs tested 200 times with +/- 10% component tolerance for all resistors and capacitors. The frequency response in the desired range (1 - 1kHz) was observed.

B. A Low Noise, Low-Power EEG Acquisition Node for Scalable Brain-Machine Interfaces




5. Conclusions
The simulations for both circuits were not successful in that the designs were either not realizable or not simulated correctly. Given more time analysis into the faultiness of the design simulations would be recommended. Temperature analysis and further Monte Carlo sensitivity analysis is possible and only useful if the nominal circuit operation can be realized.

6. References
[1] C. M. Horwitz. “Low-Drain EEG Amplifier,” IEEE Transactions on Biomedical Engineering, pages 60 – 61. Available: http://www.springerlink.com/content/r385776525771337/ [2] Thomas J. Sullivan and Stephen R. Deiss and Gert Cauwenberghs and Tzyy-Ping Jung. Department of Biology, Institute of Neural Computation, University of California, San Diego. “A Low-Noise, Low-Power EEG Acquisition Node for Scalable Brain-Machine Interfaces,” 1. Available: http://www.google.com/url?sa=t&source=web&cd=1&ved=0CBYQFjAA&url=http%3A%2F%2Fciteseerx.ist.psu.edu%2Fviewdoc%2Fdownload%3Fdoi%3D10.1.1.84.5129%26rep%3Drep1%26type%3Dpdf&rct=j&q=A%20Low-Noise%2C%20Low-Power%20EEG%20Acquisition%20Node%20for%20ScalableBrain-Machine%20Interfaces&ei=A1nyTc_IOo7ksQOanoy0Cw&usg=AFQjCNFH5T0Cgj-ID0tg4Ed8dfdKzb4azA