Yale University

PHYS 381/504: Op. amps., quartz resonators, and noise

Introduction
Johnson Noise
Macroscopic Derivation
Microscopic Derivation
Install and Test the Optical Amplifier
Install the Tuning Fork
Welcome to LabView
Tune out the Package Capacitance
Gather Data
Load the Driven Circuit Data
Fit the Driven Circuit Data
Find Circuit Characteristics from this Fit
Analyze the Undriven Circuit from Characteristics
Final Results
References
-------------------------------------------------------
PHYS 381La HOME
Classes Page
Yale Graduate School
Yale University

Andrew Pariser, December 2005
modified S.M., January 2006

© 2006 Yale University, New Haven, Connecticut 06520

Analyzing the Undriven Circuit

We determine that there is some external noise source that keeps the background noise level at −125 dB.  We add this in quadrature to the Johnson Noise coming from our tuning fork, and define this result as our noiseFunction.
The Johnson Noise of the tuning fork is given by (4kTRΔf)^(1/2)LC(f) where LC( f ) is the fitted Lorentz Curve that matches the resonance peak of this particular tuning fork with its specified RLC values, k is Boltzmann's constant, T is the temperature of the experiment and R the tuning fork's equivalent resistor.

noiseFunction = ((10^(-12.5))^2 + ((4k T R 125000)^(1/2) * dataFit)^2)^(1/2)/.{k→1.38 * 10^(-23), T→300} ;

Read in the undriven data and zoom in on the peak.

johnsonNoise_dB = ReadList['JohnsonNoiseTunedGoodRun1', {Number, Number}] ;

First we show that there is a whole lot of noise across the whole frequency band

ListPlot[johnsonNoise_dB, PlotRange→All, PlotJoined→True]

[Graphics:HTMLFiles/data_undriven_analysis_5.gif]

-Graphics -

We prune out all the data outside a small window around our regular resonance peak

johnsonNoise_dB = Drop[Drop[johnsonNoise_dB, 15850], -45600] ;

Now we show how successfully the noise prediction matches the data:

file

[Graphics:HTMLFiles/data_undriven_analysis_9.gif]

-Graphics -

Determine the maximum predicted value of our noise output

10Log[10, noiseFunction]/.f→32766

-94.7563

We then find the height of our data at this maximum, and observe that the actual measured noise value is less than the predicted.

Flatten[Select[johnsonNoise_dB, #[[1]] == 32766&]]

{32766., -107.852}         

Top