|
Basics of Instrumentation, Measurement and Analysis
Program
Meeting |
Date |
Topic |
Material
(to be populated) |
1 |
23.09 |
Introduction to electricity: Electric charge, Current and Potential. Resistance, Capacitance and Inductance. Direct and Alternating Current. Frequency. Unwanted properties, Impedance. Ohm and Kirchoff’s laws. Voltage and current measurements. Composition of unequal components: Filters. Response of high/low-pass RC-filters to the voltage step and sinusoidal wave. Signal conditioning before digitization. Soldering introduction. Exercize B: exploring characterictics of a simple passive low-pass and high-pass filters. Signals will be generated by the signal generator and analyzed with the help of oscilloscope.
Manuals: Tektronix_TDS420A_oscilloscope.pdf, Tektronix_TDS200_oscilloscope.pdf.
Correction of B & W book: Equation on p. 6 and following explanation are not correct. It should be U = L*dI/dt. For explanation see H&H p. 28. Equations on p.13 are wrong, as they don't take into account the phase shift between voltages on R and L.. The correct is Z = R+i*Omega*L, where i^2 =-1. For explanations see H&H pp. 31-32. |
|
2 |
30.09 |
NPN transistor. Emitter follower pp. 62-65 H & H. Common-emitter amplifier pp. 76-77 H & H. Classic transistor differential amplifier pp. 98-100 H & H.
Exercise C: Building operational amplifier (OA) band-pass filter for the microphone – microphone preamplifier. Test of the filter frequency response. Signal from the piezoelectric microphone consists from a constant shift from the ground about 0.9V and the alternative component in range of +/-50mV. To digitize such signal one should remove the constant component and very low frequencies (<20 Hz) from the signal by the high-pass filter. However, frequencies of most human speech harmonics lie above 100Hz, thus, we shall take the signal above this boarder. Afterwards one has to amplify it to make the output compatible with the standard audio recording equipment that has input range +/-1 V. Also, frequencies of most of harmonics in human speech lie below 10 kHz. Thus, we shall take the signal in frequency range 100 Hz - 10 kHz. We shall learn how to realize all these functions utilizing only one operational amplifier.
Datasheets: Amplifier_opa2350.pdf, Microphone_KPCM_G60H50_44DB_1184.pdf. |
|
3 |
7.10 |
8.00 - 9.00: Discussion of Exercise C. Analysis of non-stationary signals: convolution and Laplace transform. The following sources are recommended for in-depth reading:
The Laplace transform (lecture S. Boyd, Stanford).pdf
Table of Laplace Transforms (S. Boyd, Stanford).pdf
Manuscript LaplaceTransform.pdf
Manuscript Laplace_Transform(numerical approach).pdf
Correction of B & W book: In the table p. 139 for Ramp should be F(s) = c/s^2, for Step should be F(s) = c/s, for Low-pass RC filter should be h(t) = a*exp(-a*t), H(s) = a/(s+a), for Output should be G(s) = a/(s*(s+a)). The same corrections should be introduced in the scheme below this table.
9.00 - 12.00: Introduction to LabVIEW (given by Gagan): graphical programming, data acquisition, basic exercises. Loops, data format, sub vi's, saving data to files, arrays, timing.
Exercise D.
Please check out some of the videos in the "Getting
Started with LabVIEW" playlist from the LabVIEW YouTube channel: http://www.youtube.com/user/Labview?blend=7&ob=5#p/c/B968815D7BB78F9C/0/ZHNlKyYzrPE
They are an excellent introduction to some basic features of LabVIEW.
Highly recommended are:
Writing Your First LabVIEW Program
Data Flow Programming Basics
LabVIEW Data Types
Using Debugging Tools in NI LabVIEW
Using the Tools Palette in NI LabVIEW
Using Arrays in NI LabVIEW
Using LabVIEW Case Structures
Using Loops in LabVIEW |
|
4 |
14.10 |
Data acquisition and control in LabVIEW (given by Gagan). Exercise E: Record and save a sound wave file, control an LED.
Introduction to parallel loops. Queues. Exercise F. |
|
5 |
21.10 |
Burst Detector project: Discuss the overall scheme of the project. What are the aims, process components and machine learning methods we will study to classify neural bursts, and separate them from normal spiking mode, movement driven noise etc. How shall all these components come together to do real-time classification? We will development a scheme for the GUI that allow us to save data at random time points and add "tags" to them at s specific times in files. We will look at a simple example of how to use the state-machine coding example to record, display, buffer, and save data continuously. We shall try to implement the signal conditioning stages of software. |
|
6-9 |
28.10,
4.11,
11.11, 18.11 |
Theoretical basis for data classification: We will study simple linear methods to classify the data, such as Fisher Linear Discriminant Analysis, Principal Component Analysis for a better representation, and perceptrons. We will study our training dataset and check whether it is linearly separable or not. We shall also look at simple features of the data (such as spike timing differences) to see if they give us more insight. For this, a threshold and peak detection operation will need to be implemented. Following the theoretical part, students groups shall work separately on one of the algorithms to add as sub-components to the overall code.
The last 3 classes shall be time allotted to develop the program.
|
|
10 |
25.11 |
Measuring high-resistance and low-capacitance. Exercise G: Building an impedance meter for microelectrodes, testing the impedance of metallic electrodes. Knowing the impedance of electrophysiological recording electrode is important for estimation of its recording capabilities – number of cells that can be recorded or separated, sensitivity to noise and etc. However, current that one should pass through such electrode for the measurement should not be too high: cells in the vicinity of the electrode tip can be destroyed. For the extracellular recording electrode the current should not exceed 30 nA. Most of electrical activity of single cells (spikes) lie in the frequency range 300-3000 Hz. Thus, to estimate suitability of the electrode to record such activity, its impedance is usually measured at 1000 Hz. To measure the impedance of the microelectrode we shall generate 30nA sinusoidal current with the help of signal generator and current-limiting resistor. We shall estimate the impedance of the electrode by the voltage drop on it looking at the oscilloscope. Phase shift between the generator output and signal at the electrode will allow estimate resistive (active) and capacitive parts of electrode impedance separately. Preliminary measurement of voltage drop of at non-immerged in the solution (“virtual brain”) will allow to subtract the parasitic impedance of conductive wires to get precise estimate of the electrode impedance. |
|
11 |
2.12 |
Electrochemistry, electrolytes, AgCl electrode. Metal electrodes goldplating. Measuring electrode impedance before and after goldplating. Electrical processes in living organisms take place in watery solutions containing salts, proteins, carbohydrates and a host of other organic and non-organic substances. These processes are dominated to a large extent by various salts. Therefore, we will need a good understanding of the properties of electrolyte solutions and of the processes associated with them. In addition, most methods to get measurements from the wet medium are carried out with electronic instruments, which must be connected somehow to the process studied. Therefore, we are interested also in the processes at the electrodes used for measurement and stimulation. Exercise H: Goldplating of the metal electrode. Measuring its impedance before and after goldplating. Fabrication of AgCl electrode by electrochemical oxidizing the silver wire in 0.5M KCl solution (forming silver chloride layer). Measuring I/V behavior of the electrode before and after oxidizing.
SIFCO Process Gold (Alkaline) Material Safety Data Sheet: MSDS-3023.pdf. |
|
12-13 |
9.12, 16.12 |
9.00 - 12.00: Analysis of electrophysiological data in Matlab. Digital filtration. Spectral analysis. Spike sorting.
WaveClus: wave_clus_2.0wb.zip
WaveClus introduction and tutorial: wave_clus_intro.pdf, Spike_Sorting_Tutorial.ppt
Dataset: RA_16ch_25kHz_1min.mat, it's pictures: RA_300dpi.tif, GoodNeuron100ms.tif
Script for import of multichannel data in WaveClus: data_convert.m
Spike sorting methods comparison: Wild2012(spike sorting comparison).pdf
WaveClus theoretical description: Spike_sorting_WaveClus.pdf, Blatt1996(SuperparamagneticClusteringOfData).pdf.
KlustaKwik descriptions (an alternative method): Spike_sorting_KlustaKwik.pdf, KlustaKwik_environment.pdf. Example of usage KlustaKwik for sorting the same dataset: KlustaKwikExample.zip. This example works under Windows and Linux. For using on Mac small changes are needed.
Explanation of finite impulse response (FIR) filter design: Design_of_FIR_Filters.pdf. |
|
Please direct all questions to Alexei Vyssotski
Last updated July 19, 2016
|
